Here,I will discuss the planet that we live on and other bodies like it, namely the three other terrestrial planets Mercury,Venus, and Mars, and also the Moon. To the astronomers of the IAU, the Moon is just a satellite, but its composition and internal structure place it among the terrestrial planets from the perspective of a geologist or geophysicist. Figure 3 shows these five at the same scale, and Table 3 lists some relevant data. Within this group, Mercury and the Moon have effectively no atmosphere. Venus has only slightly lower size, mass, and density than the Earth, so gravity at its surface is only slightly less than on the Earth. However, its atmosphere is considerably denser. Mars is larger than Mercury but less dense. These two effects offset each other so that their surface gravities are very similar, but being colder, Mars has been able to hold on to a thin but respectable atmosphere. The Moon has the lowest surface gravity of all – about one-sixth of the Earth’s – which is why Moon-walkers bound around so strangely. Mean surface temperatures obscure wide variations with latitude and, in some cases, between day and night. For example, the hottest daytime temperature on Mercury exceeds 400 °C, whereas at dawn after a long Mercurian night the temperature is below –180 °C.

Table 3　Basic data for the terrestrial planets

3.Top: from left to right, Mercury, Venus, Earth, Moon, and Mars,shown at the same scale. Bottom: the much larger giant planetsJupiter, Saturn, Uranus, and Neptune, with the terrestrial planetsinserted to the same scale

Cores

Terrestrial planets are distinguished by having rocky exteriors,made largely of silicate minerals. However, their densities are too great for them to be rocky throughout, and it is believed that each has an iron-rich core at its centre. No planet’s core can be seen or sampled directly, but there are several independent lines of evidence. Density is one, showing that the interior must be denser than rock even allowing for internal compression at high pressure,and analyses of the trajectories of orbiting spacecraft show that density increases symmetrically about each planet’s centre.Chemical models for what is likely to happen inside a rocky planet suggest that there is insuffi cient oxygen for all the iron to be oxidized and bound up in silicate minerals. Thus, if the interior had ever been molten this would have allowed metallic iron,which is denser than rock, to sink towards the centre. This is an example of a processes called differentiation.

The outer parts of the iron-rich cores of the Earth and Mercury must be molten today, because those two planets have strong magnetic fi elds, apparently generated by dynamo motion in an electrically conducting fl uid. For such a small planet, Mercury’s density is very high, so its core must be exceptionally large,occupying about 40% of its volume and accounting for nearly 75%of its mass. Magnetic fi elds are not being generated inside Venus,the Moon, and Mars, so their cores are probably entirely solid.

In the case of the Earth, we have additional evidence about the core from studying how seismic waves, which are vibrations triggered by earthquakes (or underground nuclear tests!), travel through the planet. This confi rms a solid inner core 1,215 kilometres in radius and a fl uid outer core 3,470 kilometres in radius. Both seem to be mainly iron alloyed with 5%–10% nickel, but density arguments require something less dense than iron too, making up 6%–10% of the outer core and 2%–5% of the inner core. The likeliest explanation is some mix of oxygen, silicon, and sulfur.

In total, the Earth’s core occupies about 16% of the planet’s volume. Comparable values for Venus and Mars, which are estimates based largely on their average densities, are about 12%and 9%, respectively. There are some very limited seismic data from the Moon (from the Apollo programme), hinting at a relatively small core between about 220 and 450 kilometres in radius (less than 4% of the Moon’s total volume). About 1 in every20 meteorites is made of an alloy of iron with 4.5%–18% nickel,corresponding to the cores of planetesimals from the asteroid belt that had differentiated internally before being broken up by collisions.

Mantles and crusts

The silicate part of a terrestrial planet surrounding its core is called the mantle. This makes up the majority of the total volume of each terrestrial planet, and most of their mass except for

Mercury. The crust is a relatively minor unit overlying the mantle.It is also of silicate, though slightly different in composition to the mantle.

A planet’s present mantle evolved from the molten rock that would have covered the globe after the fi nal giant impact collision,known to geologists as a ‘magma ocean’. While a magma ocean cools, its surface must radiate heat to space, causing it to chill to a solid skin. However, this skin would continually be broken and stirred back in, thanks to turbulence below and impacts from above. The magma ocean would continue to cool, but, unlike the freezing of a ball of water, there is no discrete temperature at which the whole of it would become solid. The nature of molten silicate material is such that minerals of various compositions crystallize out at different temperatures and pressures. Planetary scientists are unsure of the extent to which magma oceans crystallized in layers, or whether minerals denser than the melt were able to sink while less-dense minerals were able to rise,perhaps sticking together to form massive ‘rock bergs’ that could force their way up more effectively.

Aggregations of this flotation material, chemically different to the underlying magma ocean, formed the earliest true crust on the Moon, where it survives today as the lunar highlands (the pale areas on the Moon’s face). On the larger terrestrial planets, the nature of the oldest crust has not been determined, partly because it has largely been replaced (or, at least, covered over) by later kinds of crust. To see how this might happen, we have to turn our attention back to the mantle. As a young planet cools, eventually its mantle becomes fully solid. Two important characteristics of silicate materials now become relevant. The fi rst is that suffi ciently hot solids are neither completely immobile nor undeformable. Hot rock in a planet’s interior is capable of fl owing at speeds of a few centimetres per year (the rate at which your fi ngernails grow), in much the same way as a block of pitch will deform over time. Within a solid mantle, motion will occur at a slow but geologically effective pace if there are any forces capable of driving it. Inside a planet, the necessary impetus comes from heat. Hotter mantle from deep down will be slightly less dense than the cooler mantle above, so there is a tendency to swap places. Motion of this sort is called convection, and is what you can observe in a pan of soup heated on a hob, except that within a planet it is much slower ‘solid state convection’.

Imagine a streamer or ‘plume’ of hot mantle welling upwards and displacing colder mantle downwards. As it gets closer to the surface, the pressure experienced by the rising mantle decreases,which brings into play the second relevant characteristic. As pressure falls, silicates begin to melt. The process is called ‘partial melting’, because only part of the solid melts, and the magma that forms is slightly richer in silica than the solid from which it was extracted. The resulting magma is also less dense than the solid,so buoyancy forces will squeeze it upwards towards the surface,especially if there are pathways where the overlying rock is under tension or fractured. Unless it stalls below ground as an intrusion,the magma will be erupted through volcanoes.

Rock formed in this way is described as igneous, and crust produced by igneous activity can replace the original crust of a planet by infi ltration or burial. The dark patches on the Moon, the lunar ‘maria’, are low-lying regions where the paler primary crust has been buried by lava fl ows produced in this way. The presentday crust of the Earth results from partial melting of the mantle to make oceanic crust, and from melting and recycling of many generations of oceanic crust to make continental crust. The Earth’s oceanic crust is 6–11 kilometres thick, whereas continental crust varies from about 25 kilometres in thin, stretched regions to90 kilometres below major mountain ranges. In total, crust occupies only about 1% of the Earth’s total volume. The Moon’s crust averages about 70 kilometres in thickness (13% of the Moon’s volume), ranging from 〉100 kilometres in some highland regions to 〈20 kilometres below some major impact basins.

In summary, crust is chemically related to the underlying mantle, but differs in ways depending on how it was extracted from it. Crust is lower in density and its average composition is richer in silica than the mantle. Crust is more varied than mantle, and includes rock that has reacted chemically with any atmosphere or liquid water, and that has been broken apart or dissolved, transported (by gravity, wind, water, or ice), and deposited elsewhere. Such deposits constitute sedimentary rock.Burial, deformation, and heating can cause sedimentary or igneous rock to recrystallize, in which case it is known as metamorphic rock.

Internal heat

Planets are hot inside partly because of heat left over from their accretion. For a bigger planet, the fraction of this ‘primordial heat’remaining today is greater. This is because heat content is related to planetary volume, which depends on the cube of the radius,whereas heat leakage is limited by surface area, which depends only on the square of the radius.

Heat is also generated inside a planet, by decay of radioactive isotopes. There are many of these, but only four whose decay produces signifi cant heating: potassium-40, uranium-238,uranium-235, and thorium-232. Because of their geochemical affi nities, these elements are more abundant in crustal rocks than in the mantle. In the Earth, approximately the same amount of heat is generated radiogenially (that is, by radioactive decay) in the crust as in the whole of the volumetrically much larger mantle.

A terrestrial planet’s total content of heat-producing elements depends on its mass (and hence its volume). Just like primordial heat, radiogenic heat is retained more effectively in larger planets.In the case of the Earth, about half the heat leaking to the surface today is primordial, and almost all the rest is radiogenic.

Lithospheres

The transition in properties from cold and rigid to warm and convective generally occurs at some depth below the boundary between crust and mantle. Thus the crust and the uppermost mantle constitute a single mechanical layer, forming a rigid outer shell. This shell is called the ‘lithosphere’, using the Greek word lithos (‘rock’) to indicate that the layer has the mechanical properties of everyday rock. Below the lithosphere is where the mantle, although rocky in composition, is hot enough and weak enough to convect. This zone is sometimes called the asthenosphere (using Greek a-sthenos meaning ‘without strength’).

The Earth’s lithosphere is about 100 kilometres thick and is fractured into a number of plates, which are able to be shunted around thanks to the special weakness of the underlying asthenosphere. As part of a process known as ‘plate tectonics’, new lithosphere is created where plates pull apart (usually hidden from view below the ocean) and destroyed where one plate is drawn below another, at subduction zones marked by trenches on the ocean fl oor. The grinding of one plate against its neighbour is the cause of most earthquakes. If anyone tells you that the Earth’s plates are ‘crust sliding over the mantle’, they are wrong, repeating a persistent fallacy that appears in many school textbooks and examination syllabuses. The truth is that a plate consists of crust and the conjoined rigid uppermost mantle, which together slide across the deeper, less rigid, asthenospheric mantle.

The lithosphere, being brittle, is the layer where faults can occur,as one mass of rock grinds past another. Faults are common on the Earth, especially in the zones where two plates meet, and can be identifi ed on other planets too ( Figure 4 ).

Plate tectonics seems to be unique to the Earth. Greater lithospheric thickness in the more easily cooled smaller bodies of Mercury, the Moon, and Mars undoubtedly contribute to this, but a more important factor is that for plates to be mobile, the top of the asthenosphere needs to be especially weak. Within the Earth, this is accomplished because of a small amount of water within the rock, which weakens it and encourages the formation of a small amount of melt that lubricates grain boundaries. Venus has lost its water, so its asthenosphere is dry and lithospheric plates cannot slide freely across it.

4.A 500-kilometre-wide view of part of Mercury. Solar illuminationis from the right. Shadow picks out a kilometre-high scarp with theshape of an open letter M on its side. This is an ancient thrust faultnamed Beagle Rupes, marking where the terrain on the right (east) hasbeen pushed westwards over the terrain on the left (west). Some of thecraters are older and others younger than this fault

A planetary asthenosphere that is dry, or very deep, is manifested principally by two effects at the surface. One is the height of mountains and depth of basins. If these are too great, the asthenosphere will fl ow and fl ex the overlying lithosphere, thereby reducing the topographic contrast until it is small enough to be supported by the strength of the lithosphere alone. The second is the pattern of fracturing caused by large impacts. An impactor several tens of kilometres in diameter arrives with suffi cient force for the resulting crater-forming shockwaves to disrupt the lithosphere, and the crater will take the form of a basin marked by concentric rings of fractures. In a thinner lithosphere, the rings tend to be closer together, so these multi-ringed impact basins can be used to estimate the depth to the asthenosphere at the time of their formation. As a planet slowly cools, its lithosphere becomes gradually thicker.

Volcanism

Magma, the name given to molten rock before it erupts, can be generated inside a planet essentially by three different causes.Direct application of heat is only one of these, and is often the least important: slow build-up of heat trapped below a planet’s lithosphere may account for some episodes of widespread volcanism, and strongly varying tidal stresses inside a planetary body work against internal friction, leading to ‘tidal heating’.Alternatively, decreasing pressure in an upwelling zone in the mantle can cause partial melting (for example, leading to the generation of Earth’s oceanic crust). Also, it is possible that sudden reduction in pressure, as must happen to the mantle below where a major impact basin is excavated, could trigger a melting event. The third mechanism is to introduce water into the mantle or lower crust. Water reduces the temperature at which silicates begin to melt. The Earth has chains of volcanoes above subduction zones, because water that has been dragged down within the rocks of the subducting plate escapes upwards into the base of the over-riding plate. Here conditions are not hot enough for melting when dry, but partial melting will begin as soon as water is introduced even though there has been no rise in temperature.

The Moon

People began to speculate about volcanism on the Moon almost as soon as craters were seen through telescopes. They were on a false trail because, as we are now quite certain, almost all the craters on the Moon were made by impacts. In fact, the major volcanic areas of the Moon are the dark patches that were once thought to be dried-up sea beds. That is not the case, though they still bear the name ‘mare’ (pronounced mah-ray), which is Latin for ‘sea’. The plural is ‘maria’ (pronounced mah-ri-a). These cover about 17% of the Moon’s surface, mostly on the near side, which is the hemisphere that permanently faces the Earth. Here, lava with a composition similar to terrestrial basalt has fl ooded the major multi-ringed impact basins.

Specifi c vents where the mare basalts were erupted are hard to identify ( Figure 5 ). Clearly, they did not take the form of conical volcanoes. Most likely, they were fi ssures through which fountains of incandescent molten lava were expelled by the force of expanding volcanic gas to heights well in excess of a kilometre. On falling to the ground, the lava was still hot enough to spread out,fl owing downhill for hundreds of kilometres. Most of the fi ssure vents sealed themselves over as their rate of eruption waned, or were buried by later eruptions.

Four of the six Apollo Moon landings (1969–72) were on maria,which are fl atter and so safer places to land than the highlands.Samples of mare basalt brought back for analysis can be dated with high precision by measuring radioactive decay products within them (radiometric dating), and the Apollo samples show a range of mare ages from 3.9 to 3.1 billion years. This long duration of volcanism put paid to the simplest volcanic explanation for the maria, which was that the volcanism was a direct product of the basin-forming impacts. Furthermore, work since the year 2000has identifi ed some patches of mare bearing suffi ciently few superimposed impact craters that they must be younger than about 1.2 billion years. Conversely, in 2007 a fragment of lunar material found on Earth as meteorite (having previously been fl ung off the Moon as ejecta from an impact crater) was found to contain fragments of basalt dated at 4.35 billion years, half a billion years before the late heavy bombardment ended. No maria of such great age remain visible, having been buried by ejecta from subsequent basin-forming impacts. So now we know that lunar volcanism began early as well as fi nishing late.

5.A 200-kilometre-wide view of the south-east edge of Mare Imbrium.The rugged terrain on the right is highland crust uplifted in part of thebasin rim. The darker, smoother area in the upper left is mare basaltsthat have fl ooded the lower-lying ground. A 1-kilometre-wide trenchnamed Hadley Rille winds from south to north across the centre of theview, and this is believed to be a pathway along which lava fl owed froma source largely obscured by shadow. Apollo 15 landed close to HadleyRille, near the middle of the image

Mercury

Mercury is much less well known than the Moon. Under half of it was imaged by NASA’s Mariner 10 in 1974–5, and the planet then remained unvisited until NASA’s MESSENGER probe began a series of fl y-bys in 2008. This revealed details suffi cient to overcome most people’s scepticism over the extent of volcanism on Mercury. For example, in Figure 4 the smooth terrain in the lower right and fi lling the 120-kilometre-diameter basin just above right of centre is now accepted as volcanic.Previous doubts were heightened by the fact that Mercury lacks the contrast in albedo (refl ected brightness) between paler highlands and darker lavas that makes the maria so obvious on the Moon. This seems to be because the minerals making up Mercury’s lavas contain much less iron than is typical of lunar(and terrestrial) basalts. Plains formed by lava probably constitute the majority of Mercury’s surface. Some of them are old enough to date back to the era of the late heavy bombardment and are densely cratered, others are younger and have fewer craters superimposed on them.

MESSENGER imaged a few volcanic vents and also curious10-kilometre-sized blotches – some bright, some dark – that may be sites of particularly young explosive eruptions. How long Mercury stayed volcanically active is likely to remain unresolved until a spacecraft achieves orbit about Mercury and records images systematically and in better detail. The fi rst chance will be when MESSENGER begins the orbital phase of its mission in2011, and if the issue remains unresolved it should be settled by the European Space Agency’s BepiColombo mission that is due to arrive at Mercury in 2020. At present, it is safe to say that extensive areas of lava were emplaced during a period covering at least 4 to 3 billion years ago, and possibly extending to within the past billion years. Such a long duration of volcanic activity on Mercury was not anticipated, and may result from the same mysterious heat source that keeps part of its core molten.

Venus

Venus is much bigger than Mercury. Its size and mass would suggest almost as much radiogenic heat production as the Earth,and hence a similar level of volcanic activity. However, because Venus lacks plate tectonics, its volcanism operates rather differently.

Venus has a dense, permanently cloudy atmosphere, which kept the surface a total mystery until it became possible to study it by means of radar. Figure 6 shows a radar image of part of Venus obtained by NASA’s Magellan probe, which mapped almost the entire planet between 1990 and 1994. Radar images are assembled by complex analysis of the echoes bounced back in response to a continual string of radar pulses beamed at the surface. For most purposes, you can treat radar images like the black-and-white optical images that they resemble, though in fact the brightness of each feature is controlled mainly by how rough the local surface is, rather than its albedo in visible light.

Figure 6 typifi es much of Venus. It shows numerous individual lava fl ows – some rougher (brighter) and some smoother(darker) – that fl owed from west to east across the image. The lobate pattern of the individual fl ows closely resembles that of lava fl ows on Earth and Mars, but which is hard to discern on the Moon and Mercury where the fl ow margins have become degraded by impacts.

As well as lava fl ows covering about half its surface, Venus has many clearly identifi able volcanoes. Figure 7 shows an example. In the background is a 5-kilometre-high volcano with gently sloping fl anks of the kind known on Earth as a ‘shield volcano’ that results from repeated eruption of basalt through a single vent. Some individual lava fl ows can be made out on the fl anks. No one is sure how long ago this volcano and others like it last erupted. There have been intriguing hints but no proof of recent or present-day activity on volcanoes such as this, and they are rather too small for reliable crater-counting statistics. This particular volcano is built upon an older terrain unit that is smoother, except for numerous fractures upon it. The impact crater in the foreground is probably unrelated to radar-bright lava fl ows immediately to its left.

6.A 500-kilometre-wide Magellan image of part of Venus. The area ismostly lava, fed from a source 300 kilometres west of the image, but inthe south-east corner is some rugged terrain representing the oldestsurviving crust on Venus. Running from north to south in the west ofthe image is a mountain belt of ridged and fractured terrain that isbreached by the lava fl ows

Circular or elliptical patterns of concentric fractures termed‘coronae’, of which more than 300 have been identifi ed on Venus,are not thought to share a common origin with the multi-ringed impact basins of the Moon and Mercury. They can be anything from 200 metres to over 2,000 kilometres across, and are usually associated with some form of volcanism. Probably each corona marks a site where an upwelling plume in the asthenospheric mantle impinged on the base of the lithosphere. Coronae where the plume is still present are uplifted as very broad domes, whereas older ones, no longer supported by a mantle plume, have sagged. The sagging, in particular, explains the concentric fractures.

7. Computer-generated three-dimensional perspective view showing the volcano Maat Mons on Venus. This was made by draping a radar image over a model of the topography obtained by radar altimetry. Vertical scale is exaggerated tenfold. Both sets of data were collected by the　Magellan orbiter. The impact crater in the right foreground is 23 kilometres across

Impact craters on Venus are more common than on the Earth, but considerably less abundant than on the Moon and Mercury (you will not fi nd any in　Figure　6 ). There are two factors at play here. Craters less than about 3 kilometres across are entirely absent from Venus, because its dense atmosphere shields the surface from small impactors. However, larger craters are formed by objects carrying too much energy to be affected by the atmosphere. Their lack of abundance has to be explained by the young age of the surface, which works out on average to be between about 500 million and 700 million years. There are no large tracts of terrain that seem to be very much older or very much younger than the global average.

The standard explanation for this in the 1990s was that pretty much the whole planet was resurfaced in an orgy of volcanism that began 500–700 million years ago, and lasted no more than a few tens of millions of years. This would be consistent with Venus’s lack of plate tectonics leading to most heat from the deeper mantle being trapped below the lithospheric lid, until much of the uppermost asthenosphere had melted. Eventually, the cold, dense lithosphere would founder, and the buoyant magma from below would erupt. Something similar could have occurred half a dozen times since Venus was formed, and maybe it could happen again within the next 100 million years.

This model calling for catastrophic global volcanism has recently been challenged, on the grounds that the cratering statistics do not rule out a more gradual process. Progressively smaller areas could have been resurfaced by lavas at random intervals throughout the past half billion years.

Earth

On Earth, volcanism and plate tectonics combine to regulate the internal heat budget and thus prevent major asthenospheric temperature excursions of the kind postulated for Venus. Only about one-third of the heat generated below the lithosphere leaks out by conduction. Most of the heat is conveyed to the top of the lithosphere by eruption at mid-ocean ridges (where new material is added to diverging plates) and, to a smaller extent, by eruptions at volcanoes above subduction zones and at various ‘hot spots’ above mantle plumes. The asthenosphere is cooled by the reincorporation into it of the old, cold parts of lithospheric plates at subduction zones.

The closest we get to a Venus-like volcanic catastrophe is when, every few tens of million years, a region maybe a thousand kilometres across is buried by the eruption of up to ten cubic kilometres of basalt lava. This is known as a ‘fl ood basalt’. The ‘Deccan Traps’ of northwest India (66 million years), the Brito-Arctic fl ood basalts(Greenland and the north-western British Isles, 57 million years),and the Colombia River fl ood basalts (north-western USA, 16 million years) are among the better-known examples. These major but rare events could be capable of injecting so much volcanic gas, notably sulfur dioxide, plus fi ne fragments of volcanic rock known as ‘ash’,into the atmosphere that global climate could be severely affected.Figure 8 shows an example of lava fl ows on the Earth, for comparison with the images from other planets.

The way in which volcanism on the Earth probably differs most from the other planets is that expansion of gas within rising magma tends to make a substantial proportion of eruptions explosive in nature. This is for two reasons. The first is that recycled water, carbon dioxide, and sulfur dioxide escaping upwards above subduction zones adds greatly to the leakage from the deeper interior of primordial gases, so the Earth has more gas available to drive eruptions explosively. The second is that the existence of continental crust facilitates generation of magma with a higher silica content than basalt. These silicarich magmas are more viscous than basalt, so they fragment more easily. Classic ‘picture-book’ steeply conical volcanoes such as Mount Fuji in Japan are rare except on Earth, because they are symptomatic of relatively silica-rich and partly explosive eruptions.

8.A 70-kilometre-wide view from space showing the ‘Craters of theMoon’ lava fi eld in Idaho, USA. The source of the fl ows was a series offi ssures near the edge of the rugged highlands in the north-west.Compare the lobate form of the lava fl ows with the fl ows on Venus inFigure 6

Mars

Compared to the Earth and Venus, Mars has relatively few volcanoes. However, their small numbers are compensated by large size. Major groupings of large basaltic shield volcanoes occur in the Tharsis region (much of which is included in Figure9 ) and the Elysium region. Olympus Mons is the largest Tharsis volcano, measuring about 600 kilometres across its base and 24kilometres from top to bottom, which makes it the largest volcano in the entire Solar System. There are two reasons why Mars has such big volcanoes. The fi rst is that Mars is a ‘one plate planet’. Its lithosphere is an intact shell (a single plate)effectively stationary with respect to the underlying mantle asthenosphere. Unlike the Earth, where plates drift around relative to mantle plumes so that plume-fed volcanoes are carried away and cut off from their magma supply after only a few million years, a mantle plume on Mars supplies magma to the same spot in the lithosphere for as long as the plume remains active. Olympus Mons may have begun to be constructed more than a billion years ago. We have no way to tell, because we can only date (by crater counting) what is exposed at the surface today, and can’t see the older, buried,interior of the edifi ce. There are several overlapping calderas at its summit, whose fl oors are dated at around 100 to 200 million years, but the youngest lava fl ows on the fl anks appear to be only about 2 million years old, and it is likely that Olympus Mons will erupt again one day. Other volcanoes in the Tharsis region are defi nitely older, and are probably now extinct.

9.A 3,000-kilometre-wide mosaic of images showing severalenormous shield volcanoes on Mars. On the left is Olympus Mons, theSolar System’s largest volcano. On the right-hand edge is TharsisTholus, and running north-east from the centre of the southern edge,a line of three: Pavonis Mons, Ascraeus Mons, and Ceraunius Tholus

The second reason why Mars has such big volcanoes is ‘because it can’. It has a cold, strong lithosphere, about twice as thick as the Earth’s. If you transplanted Olympus Mons to the Earth or Venus,their relatively thin lithospheres would sag beneath the load, and the volcano would lose height.

High-resolution images reveal details of lava fl ows on the plains between the large volcanoes and in several other regions of Mars.However, there are also features regarded by some as volcanic that have aroused considerable controversy. Figure 10 shows a notable example.Over 30 fragments of impact ejecta from Mars have been collected on Earth as meteorites. They are all either basaltic lava or more coarsely crystalline intrusive equivalents, spanning a range of crystallization ages extending from 4.5 billion years to as young as160 million years. We can infer that igneous rock makes up the bulk of the Martian crust at depth, even though large tracts of surface have a veneer of sediment of various kinds.

10.A 50-kilometre-wide image of a controversial area of Marsobtained by ESA’s Mars Express orbiter. Some say the platy surface isa lava fl ow with a fractured cooling crust. Others see this as brokenpack ice (now dust-covered) on the surface of a frozen sea. The twoimpact craters are older than the platy surface, and their rims werehigh enough to prevent their interiors from fl ooding. The craters areactually circular in outline, but are foreshortened in this oblique view

Surface processes

Regolith and space weathering

Volcanism is driven from inside a planet, but planetary landscapes can be sculpted just as much by processes that occur essentially at the surface. On a body that is airless and therefore unprotected from space, by far the dominant process acting directly on the surface is bombardment by meteorites and micrometeorites. Fragmented material (‘ejecta’) thrown out from craters blankets the surface to a depth of several metres, and sites where solid bedrock is visible at the surface are rare ( Figure 11 ). The lunar soil, known as ‘regolith’, in which the Apollo astronauts left their footprints is composed of grains mostly a fraction of a millimetre in size, comprising fragments of crystal, tiny bits of rock, and glassy spherules that are frozen droplets of melt generated by the heat of the impact. Regolith is continually rearranged on a variety of scales by excavation of craters and dispersal of ejecta, a process known as ‘impact gardening’. On Mercury, where impact speeds are faster, regolith grain-size is expected to be about one-third that of lunar regolith.If there is no atmosphere, solar ultraviolet light can reach the surface, where it may, over time, break chemical bonds.Micrometeorite impacts and (if there is no magnetic fi eld) charged particles from the solar wind can also affect surface chemistry, so airless bodies experience a suite of processes, collectively described as ‘space weathering’, that slowly alter the composition of the surface. For example, the bonds linking iron to oxygen atoms can be broken, allowing oxygen to escape and leaving submicroscopic grains of pure metal, called ‘nanophase iron’.

11.Telephoto view looking across Hadley Rille, taken by Apolloastronaut Dave Scott. The 2-metre-thick horizontal layer runningacross from the left is a rare example of bedrock (probably a lava fl ow),here exposed on a steep slope. Everywhere else is covered by regolithranging in size from boulders down to dust

When a planet has an atmosphere, only the largest, and infrequent, impactors can reach the surface at high speed. For example, in the Earth’s atmosphere, stony asteroids less than about 150 metres in size are likely to be sheared apart. The resulting fragments are small enough to be slowed down by friction, so by the time they reach the ground they have lost almost all their initial velocity, and do not form craters. Meteoritic dust, which is mostly micrometeorites but also fragments frictionally ablated from larger meteorites, settles to the ground at an average accumulation rate of 0.1–1 mm per million years. This dust makes such a tiny contribution to the total rate of sedimentation that it is totally swamped by other sediment, except on deep ocean fl oor far from land.

Erosion and transport

Impact gardening aside, processes that can wear away rock and transport the resulting fragments are wind, fl owing water, and moving ice (glaciers). Water can also dissolve rocks, during chemical weathering. Elements carried in solution by water may reappear elsewhere, being precipitated in new minerals. This applies especially to salt deposits, and also to many forms of carbonate rocks. However, on Earth most limestone (calcium carbonate) is formed from fragments of the shells of marine organisms, demonstrating an important biological step in turning dissolved carbonate (or dissolved carbon dioxide gas) into solid material that can become rock.

The dust storms on Mars are famous, having been fi rst noted by telescope in 1809. At perihelion, when Mars receives 40% more solar energy than at aphelion, winds in excess of 20 metres per second can lift so much dust high into the sky that most of the surface is obscured for several weeks. Sometimes little can be seen poking through other than the summit of Olympus Mons. Because of the clouds that often congregate there, this often looks white,hence its former name of Nyx Olympica (Snows of Olympus),which was revised when images from spacecraft showed what was really going on.Many signs of the action of wind on Mars can be seen from orbit or on the ground ( Figure 12 ) in the form of sand dunes and also smaller wind-ripples in the surface dust. Some of the dunes on Mars are being actively sculpted by the wind, but others have probably not changed their form for millions of years. Wind-blown sand is a powerful agent of erosion on Mars. The low atmospheric density means that a wind capable of transporting sand grains has to be blowing much faster than on Earth, and some exposed layers of rock have become curiously sculpted by abrasion.

12.Some sand dunes, needing only camels or a palm tree for scale. Infact, this picture was taken by NASA’s Opportunity rover on the surfaceof Mars, looking obliquely down from the rim of a crater onto a fi eld ofdunes on the crater fl oor. The visible area is about 100 metres across

Venus has a much denser atmosphere than Earth, with 92 times greater ground-level atmospheric pressure. Even sluggish winds can shift sand particles around, and Venus has several fi elds of sand dunes. However, here when a wind-blown grain strikes an exposure of bedrock its erosive power is limited, partly because of the slow speed and cushioning of the blow by the dense air, and partly because the high surface temperature of 480°C makes material deform plastically rather than abrading in a brittle fashion.

For Earthlings, fl owing water is usually the most familiar agent for transporting sediment – in a river, or as waves on a beach. In the Solar System, nowhere other than Earth currently has surface conditions allowing liquid water to be stable. Venus is far too hot,and although the noontime temperature on Mars can creep well above 0 °C, its atmosphere is so tenuous that ice at the surface turns directly to vapour rather than melting. However, there is abundant evidence that water once fl owed in prodigious volumes across the surface of Mars ( Figure 13 ). Mars has suffered extremes of climate at least the equal of Earth’s, and billions of years ago its atmosphere was suffi ciently dense and wet to permit rainfall and catastrophic fl ooding. The largest canyon system in the Solar System, Valles Marineris (‘the Valleys of the Mariner’, named after being discovered on images returned by the probe Mariner 9 in1971), is a 4,000-kilometre-long rift system initiated by fracturing of the crust, but widened by erosion when water fl owed through it.

13.A series of east–west fractures attests to the tectonic origin ofMars’s Valles Marineris canyon complex, of which only a fraction iscovered by this 800-kilometre-wide view. Note the winding and deeplyincised channels feeding into it from the south, which show the roleplayed by fl owing water in widening the main canyon

At its deepest point, the fl oor is 7 kilometres below the rim(Earth’s Grand Canyon in Arizona is only 2 kilometres deep), and it is so wide that if you stood on one rim, the opposite side would be out of sight beyond the horizon.

Despite its vastness, Valles Marineris was not recognized by pre-Space Age telescopic observers. The notorious ‘canals’ of Mars mapped by the Italian Giovanni Schiaparelli in 1877, and subsequently championed by the American Percival Lowell who until his death in 1916 thought they were giant works of engineering by intelligent Martians, are illusory. They bear no relation to any of the many genuine channels on Mars. Of these,examples fed by a branching network of tributaries (including many much longer versions of those shown in Figure 13 ) are likely to have been supplied by rainfall. The water fl owing along others probably leaked out of the ground, possibly when permafrost melted. The streamlined shapes of the ‘islands’ where channels debouch onto the plains show them to have been scoured by catastrophic fl oods. Robotic landers ( Viking 1 in 1976 and Mars Pathfi nder in 1997) that touched down in such places found an abundance of rocks dumped there by fl oodwaters.

All the major valleys on Mars have numerous impact craters superimposed upon them, so clearly they must be ancient, having last fl owed well over a billion years ago. Since then, many have suffered landslips from their walls, and their fl oors now bear trains of sand dunes formed by the cold wind howling along their length. In the 1970s and 1980s, most scientists would have told you that although Mars experienced at least one wet epoch in its distant past, it is now extremely dry except at the poles where there are small caps of water-ice. Imagine everyone’s surprise when in 1999 a high-resolution imager called the Mars Orbiter Camera began to reveal gullies only a few metres wide and a few hundred metres long on steep slopes in several places on Mars.The lack of superimposed craters and the observation that in many cases debris fans at their downstream ends had begun to bury sand dunes showed they must be young features, but how young? Proof was not long in coming that some are still active today, when repeat images began to reveal changes ( Figure 14 ).

The debate has now moved on from questioning the age of the youngest gullies, and now focuses on how they are cut. One theory is that water is responsible. There could be reservoirs of liquid groundwater under pressure in the Martian subsoil. Where a slope,such as the crater wall in Figure 14 , cuts below the water table, a barrier of ice within the soil (‘permafrost’) would normally prevent its escape. However, if the barrier were temporarily to give way,water could come gushing out. The liquid would not be stable – it would be both boiling and freezing as it fl owed – but it could traverse the length of one of these gullies before completely evaporating. Sceptics argue that liquid fl ow is not necessary to carve the gullies, and that they can be explained as a result of dry rock avalanches.

14.Two views of the same 1.5-kilometre-wide area covering the innerwall of a 6-kilometre-diameter Martian crater, recorded in August 1999(left) and September 2005 (right). The rim cuts across the top left, andthe fl oor is towards the lower right. There are many gullies eroded intothe slope of inner wall, one of which seems to have fl owed betweenthese two dates, carrying some pale debris onto the lower slope

Some Mars scientists see evidence for glaciers, especially at the eroded edges of highland plateaus. There is no ice exposed at the surface today (except that at the poles), but the rock-strewn ground revealed in high-resolution images from orbit could be debris covering (and insulating) underlying ice. Ground-penetrating radar data obtained from Mars orbit lend credence to this, which is one of the reasons why I am happier to accept the region shown in Figure 10 as a dust-covered frozen sea rather than as a lava fl ow.

Channels on the Moon such as Hadley Rille ( Figure 5 ) were lava pathways and were certainly not cut by water, and the only lunar water is small quantities of ice in the regolith near the poles. More than 200 sinuous channels have been mapped on Venus, one of which is 6,800 kilometres long. It is very unlikely that Venus experienced suffi ciently extreme climate change for liquid water to have existed recently enough to erode these channels, so they too were probably cut by lava.

Naming surface features

I have used names of surface features on other planets several times already: Olympus Mons, Valles Marineris, Hadley Rille, and so on. Without such names, I would be reduced to referring to them as ‘the biggest volcano on Mars’, ‘Mars’s giant canyon system’, and ‘that big trench near where Apollo 15 landed’. Less notable features would be even harder to describe, unless by a totally unmemorable coordinate system.

But nobody lives there, so who allocates names and how offi cial are they? When astronomers fi rst started to draw maps through their telescopes, some were suffi ciently independently minded to invent their own names, often regardless of any previous work.An early task of the IAU (founded in 1919) was to sort out the mess, arrive at single offi cial names for multiply named features,and to establish standards and conventions for allocating future names. This applied to the names of newly discovered bodies and also features on the surfaces of planetary bodies that it might become desirable to name or that would become visible thanks to improved imaging techniques. Originally, the latter meant simply bigger and better telescopes, and few founders of the IAU can have realized that they had established a means for supervising the nomenclature of features revealed by visiting spacecraft.

Some have castigated the IAU’s handling of Pluto’s reclassifi cation,but I know of no one who thinks badly of the basis for the IAU-administered naming process. This is fair, non-political, and seeks to represent all the world’s cultures – not necessarily on any single body, but balanced across the whole Solar System.

Building on what had already become common practice for lunar features, IAU nomenclature assigns a single unqualifi ed name to craters, whereas most other features are given a name plus a Latin descriptor term that denotes what kind of a feature it is. Thus‘Olympus Mons’ means ‘Olympus Mountain’, telling you immediately that this feature is a mountain named Olympus. Note that although no one doubts that Olympus Mons is a volcano, the descriptor term does not say this. Descriptor terms intentionally avoid interpretation (which may turn out to be wrong) and stick to description .

Common descriptor terms that you may meet are: chasma (a deep, elongated, steep-sided depression), fl uctus (a fl ow-like feature), fossa (long, narrow, shallow depression), mensa (fl attopped prominence with cliff-like edges), planitia (low-lying plain), planum (high plain or plateau), rupes (scarp), and vallis(branching valley). On the Moon, there is also mare (plural maria)which translates as ‘sea’ but had become too entrenched to be replaced by a more apt term.

There are also themes for names on each planet. Lunar craters are named after famous deceased scientists, scholars, and artists,whereas maria take Latin terms describing various weather conditions. Mars is the only place other than the Moon with signifi cant heritage of names from before the IAU became involved. These, with modern descriptor terms added, come from telescopic mapping by Giovanni Schiaparelli and Eugenios Antoniadi in the late 19th century, and mostly refer to broad regions such as Tharsis and Elysium. Each large valley is given the name for Mars in a different language, whereas small valleys are named after rivers on Earth. On Venus, almost all names are female: craters are named after famous historical women and most other features after goddesses. On Mercury, craters are named after dead artists, musicians, painters, and authors,whereas scarps (rupes) are named after scientifi c expeditions or the ships that carried them. Beagle Rupes ( Figure 4 ) is named after HMS Beagle on which Charles Darwin voyaged while amassing the observations that inspired his theory of evolution.

Similar principles apply to names of asteroids and the satellites of other planets. For example, Jupiter’s satellite Europa has craters named after Celtic gods and heroes, and most other features have names taken from the classical myth about Europa.

Atmospheres

After its birth, each terrestrial planet must have developed an atmosphere when internal gases leaked out via the magma ocean.These primitive atmospheres do not survive today, though the gases escaping from volcanoes show what they may have been like.The Moon and Mercury have too little gravity to hang on to a gas blanket, and the ‘atmospheres’ that you can sometimes fi nd quoted for them, with pressures much less than a billionth of the Earth’s atmospheric pressure, consist largely of stray atoms knocked of the surface by micrometeorite and cosmic ray impact.So sparse are these atoms that each is more likely to drift off into space rather than collide with another atom. This condition defi nes the ‘exosphere’ of a planet. It is merely the tenuous outermost zone of most atmospheres, but is all that the Moon and Mercury can muster.

The stronger gravity of the more massive terrestrial planets enables them to hang on to gas more effectively, although the density and chemical composition have evolved out of all recognition as a result of numerous processes. In the early days,the more active solar wind may have stripped away most of each original atmosphere, but these would be replenished by volcanism. An important ongoing process is that shortwavelength solar ultraviolet light can split molecules of water vapour into hydrogen and oxygen. Hydrogen is very light, and can escape to space, which makes this ‘photodissociation’ of water an irreversible process. Venus and Mars have both lost much of their original water in this way. The compositions of the present-day atmospheres of Venus, Earth, and Mars are summarized in Table 4 .

Having been split by ultraviolet light, atmospheric molecules can combine with others, by series of reactions described as‘photochemistry’. This occurs especially in a planet’s‘thermosphere’ which begins about 100 kilometres above the surface, so named because this layer is heated by the solar ultraviolet energy used in either splitting molecules apart or stripping them of some of their electrons. The latter process is called ionization, and ions (mainly of oxygen for the Earth and carbon dioxide for Venus and Mars) can be suffi ciently common in the outer reaches of a thermosphere to make an electrically conducting layer referred to as the ‘ionosphere’. When a solar storm brings plasma from the Sun to the Earth, this distorts the magnetic fi eld and causes unusual currents to fl ow in the ionosphere that can badly disrupt radio communications and even cause power failures.

Table 4　The present-day atmospheres of terrestrial planets,showing the abundances of the six most common gasesexpressed as a percentage of the total number of molecules(water is very variable in Earth’s atmosphere), and the surfacepressure relative to Earth

Deeper layers of an atmosphere, where the short-wavelength ultraviolet does not penetrate, are immune from photochemistry.Here the air is warmed mostly by contact with the ground (which is warmed directly by the Sun), and so in the lowest layer, called the troposphere, atmospheric temperature decreases upwards.Atmospheric pressure and density also decrease upwards, which means that the troposphere contains most of the mass of the atmosphere. In the troposphere, composition can evolve because of chemical reactions between air and rock (this is a corollary of chemical weathering) and especially (perhaps only) in the case of the Earth because of life. Here plants and primitive single-celled organisms use solar energy and atmospheric carbon dioxide to build their bodies, releasing gaseous oxygen that was vanishingly rare in the original atmosphere. Without plants, oxygenbreathing animals (like us) could not exist. The temperature would also be different, as I will explain shortly.

When air near the base of the troposphere is heated, it must expand, which makes it buoyant. It will then rise, to be replaced by colder air displaced from above. This is another example of convection (which you previously met within a planet’s mantle),and is what drives the weather on the Earth, Venus, and Mars.The pattern of circulation is different in each case because it also depends on such factors as the planet’s rate of rotation(slow for Venus), the rate of rotation of the atmosphere (much faster than that of the planet itself in the case of Venus’s upper troposphere), and the day–night temperature difference (high for Mars, small for Venus). Figure 15 shows the characteristic circulation above Venus’s south pole. In contrast, spiral storm systems in the Earth’s atmosphere tend to begin near the tropics.

The Earth’s atmosphere also differs from its neighbours in the complexity of its layering. At Venus and Mars, temperature decreases rapidly with altitude in the troposphere, then decreases more slowly with altitude in a (non-convecting) layer called the mesosphere, and then rises with altitude in the thermosphere because of the ultraviolet absorption. The Earth is unique among terrestrial planets in having a layer extending from about 10 to 50 kilometres altitude, between its troposphere and mesosphere, where temperature increases with altitude. This is the stratosphere, which is warmed by the absorption of 230–350-nanometre-wavelength ultraviolet photons (to which the thermosphere and mesosphere are transparent) by ozone molecules. Ozone is three oxygen atoms bound in a molecule (O 3 ), as opposed to two oxygen atoms (O 2 )which is what is usually meant when referring to ‘oxygen’, and is assembled from oxygen by photochemical reactions higher in the atmosphere.

15.The ‘eye’ of Venus’s 2,000-kilometre-diameter south polar vortex,imaged 24 hours apart. The dot indicates the south pole. These images,recorded in the mid-infrared, see the cloud-tops about 60 kilometresabove the surface. The centre of the ‘eye’ is warmer (appearingbrighter), showing that here the clouds are drawn downwards towarmer, deeper levels

Greenhouse effects and the hole in the ozone layer

Many people are aware of the ‘hole in the ozone layer’ and the‘greenhouse effect’, but tend to confl ate them as twin culprits of climate change. However, the two are very different.The ozone layer occurs (only) in the Earth’s stratosphere, and is where 230–350-nanometre ultraviolet light is absorbed. This is very important for ourselves and other surface-dwelling life,because if it is not blocked out such radiation can cause skin cancers and genetic damage. It takes surprisingly little ozone to be an effective screen. If you gathered all the ozone that is dispersed in the stratosphere into a single layer at sea-level it would be only about 3 millimetres thick. This is a fragile veil, so when in the 1970s and 1980s it became apparent that over Antarctica the stratosphere had lost perhaps half its ozone, there was considerable concern, and talk of a ‘hole in the ozone layer’.The main cause was traced to reactions involving industrial chemicals called chlorofl uorocarbons (CFCs) which, as a consequence, have now been banned from their former use in aerosol sprays and refrigerants so they cannot leak into the atmosphere. The Antarctic ‘ozone hole’ and a lesser one over the Arctic have now stabilized. Depletion of ozone is only a few per cent outside of the polar regions, and is undetectable over the tropics.

There is no simple link between ozone concentration and mean global temperature. A badly depleted ozone layer would make life unpleasant, but has little to do with climate change or global warming. The tropospheric temperature of a planet is controlled by how effectively the lower atmosphere absorbs infrared radiation. This is because visible sunlight warms the ground, and the warmed ground emits infrared radiation. The temperature of the atmosphere depends on two factors: the heat it picks up from contact with the ground, and how much of the outgoing infrared radiation it can absorb.

Most gas species are transparent to infrared radiation, but molecules consisting of two or more different elements absorb infrared quite strongly. Thus nitrogen (N 2 ), oxygen (O 2 ), and argon (Ar) do not absorb infrared, but water vapour (H 2 O),carbon dioxide (CO 2 ), sulfur dioxide (SO 2 ), and methane (CH 4 ) do.By analogy with trapping warmth inside a greenhouse, this is called the ‘greenhouse effect’. There is a natural greenhouse effect in the atmospheres of Venus, Earth, and Mars. Thanks mostly to its enormous load of carbon dioxide, the atmospheric greenhouse effect on Venus maintains its surface temperature an impressive500 °C above what it would otherwise be. Water vapour and carbon dioxide warm the Earth by about 30 °C, and greenhouse warming of Mars, which has a tenuous carbon dioxide-rich atmosphere, is only about 6 °C.

Earth’s greenhouse effect keeps the temperature within a range to suit the life that has evolved here. Mediated by life itself, the strength of the greenhouse effect has changed to keep the temperature in the right range. Four billion years ago, the Sun was only about 70% as brilliant as it is today, so the Earth would have been very much colder, had its atmosphere been the same as today’s. However, before 4 billion years ago, it was probably mostly carbon dioxide, and 100 times denser than today, so the greenhouse effect would have been stronger. Thanks to primitive algae, the carbon dioxide content had decreased to about 10 times its present value by about half a billion years ago, so of course the greenhouse effect must have declined too. Free oxygen (O 2 ) fi rst appeared some time between about 2.7 and 2.2 billon years ago,and peaked at about 170% of its present concentration between250 and 200 million years ago. Clearly, life on Earth has both infl uenced and benefi ted from changes in the composition of the atmosphere.

Since the beginning of the industrial era, human activity has affected the atmosphere in various ways: ozone depletion, industrial smog, and so on. However, what should concern us most is our release of carbon dioxide into the atmosphere, or, rather, back into the atmosphere, for most of this is carbon dioxide previously extracted from the atmosphere by organisms and sequestered as coal or oil. The amount of atmospheric carbon dioxide increased by about 20% in the 50 years since 1960 (faster than any natural process), and is still increasing. This ‘anthropogenic greenhouse effect’ will inevitably lead to warming of the global climate. A few degrees’ rise in temperature will affect ecosystems and will also tend to make local weather (including short-term temperature fl uctuations) more extreme. Another consequence will be a rise in global sea-level, chiefl y because water expands as it warms. So,although the natural greenhouse effect in our atmosphere is a good thing, human-induced rapid increases in the effect could have potentially disastrous consequences for our civilization.

Against a background of general gradual decrease in the natural greenhouse effect, which counterbalanced the slow waxing of the Sun’s luminosity, there have been several excursions in the Earth’s climate. Ice Ages, when much (in extreme cases all) of the surface water was frozen, are the best-known example. These are controlled not so much by the atmosphere as by variations in the tilt of the axis and the eccentricity of the orbit. Similar effects probably explain the drastic changes in the wetness of Mars’s surface over time.

Clouds

Clouds are highly refl ective, so the cloudier an atmosphere, the more solar energy is refl ected directly back into space. However, a cloudy sky increases the ability of an atmosphere to trap heat from the sunlight that does reach the ground, so the effect of clouds on global temperature is complex. The unbroken clouds of Venus have not saved its surface from being thoroughly cooked by the greenhouse effect.

Clouds form when the temperature and pressure make it favourable for some constituent of the atmosphere to condense as liquid droplets or ice particles. In the case of the terrestrial planets, the relevant constituent is usually water. Although water is only a small fraction of Venus’s atmosphere, there is enough to form a continuous layer of cloud at the top of its troposphere, between about 45 and 65 kilometres above the surface. There, water vapour condenses as droplets about2 micrometres across. These remain suspended, being too small to fall, and are described as aerosol droplets.Atmospheric sulfur dioxide dissolves in them, so they turn into sulfuric acid. However, if anyone tries to tell you that it ‘rains sulfuric acid on Venus’, they are wrong. Wherever the droplets are drawn down below about 45 kilometres by atmospheric circulation, the heat causes them to evaporate again, and they never have the chance to become raindrops large enough to fall groundwards.

Above about 6 kilometres, Earth’s clouds consist mostly of tiny ice particles, and below that altitude they are mostly droplets of water. Rainclouds are not really grey, they just look that way because they are thick enough to blot out so much light. On Mars,clouds are comparatively rare. In most of its troposphere, they are made of water-ice, whereas around 80 kilometres, near the troposphere/mesosphere boundary, clouds of carbon dioxide particles have been observed.

Polar caps and oceans

As well as condensing to form clouds, atmospheric constituents may condense as either ice or liquid at the surface. The Earth is the only terrestrial planet to have oceans today, which of course are made of water. Near the poles, water is frozen to form polar caps.The young planet Venus may have enjoyed a brief epoch when oceans covered the globe, before the evaporated water vapour(subsequently lost by photodissociation) added to a burgeoning greenhouse effect leading to the current parched situation.

However, Mars is different. A vast ‘Oceanus Borealis’ occupying the whole of the low-lying northern plains about 3.8 billion years ago was in vogue in the 1990s. Although that remains contentious,many would accept the likelihood of lakes on Mars extensive enough to be called ‘seas’ at the time when channels such as those in Figure 13 were fl owing, and some frozen relics may even survive, buried by dust ( Figure 10 ). However, there is no doubt that ice exists at the surface today in Mars’s polar caps ( Figure 16 ).These consist of ‘permanent’ water-ice with a fringe of carbon dioxide frost that grows and contracts seasonally.

16.1,500-kilometre-wide images of Mars’s northern polar cap in earlyspring (left) and high summer (right). In summer, most of the carbondioxide frost has sublimed (turned from ice to vapour), leaving onlythe residual, ‘permanent’ cap of water-ice

Earth’s and Mars’s polar caps interact with the atmosphere. They are, in effect, deposits of gases that have ‘frozen out’ of the atmosphere, either falling from the clouds as snow or condensing directly onto the ground. When the temperature rises, material from the polar caps is returned to the atmosphere, either by melting and then evaporation (for water on Earth, and probably Mars in the past) or by subliming directly from ice to vapour (on Mars, for carbon dioxide and water today).

Equilibria like these cannot occur on airless bodies like the Moon and Mercury, and so polar caps are not to be expected. However,during the 1990s it was noted that radar signals are refl ected with unusual strength from permanently shadowed regions inside craters near the poles of both bodies. This would be consistent with water-ice dispersed as grains within the regolith. A possible explanation is that the fl oors of these craters are so cold that any stray water molecules that wander into them tend to adhere to the surface in ‘cold traps’. This water need not be part of these bodies’original inventory, it could have been supplied later by impacting comets. Finding a supply of water on the Moon is of great importance if a human colony, or even just a permanently occupied base, is ever to be established there. The poles are clearly the best bet, and in 2009 water was confi rmed in an ejecta plume created when a spacecraft was crashed into a permanently shadowed polar crater. Infrared spectra obtained by other spacecraft revealed water and hydrated minerals dispersed in the regolith across broader regions, in minute concentrations but raising hopes that the Moon might not be so wholly inhospitable as formerly believed.

Cycles

Interplay between interior, surface, and atmosphere, and the cycling of components between them, is extremely important. The Earth’s ‘hydrologic cycle’ is the most familiar example. It is not a single cycle, but an array of interconnected loops. In essence,water in the oceans evaporates to form clouds, and later precipitates out as rain or snow that eventually fi nds its way back into the oceans (via riv...