As BepiColombo prepares for launch to Mercury later this year, Paul Byrne examines what the Mariner 10 and MESSENGER missions revealed about the smallest terrestrial planet. We live in a golden age of planetary exploration. With the fly-past of Pluto by the New Horizons mission in July 2015, all the major elements of the solar system – the Sun and nine “classical” planets – have now been visited. But that same year also saw the Dawn mission reach Ceres, the largest body in the main asteroid belt, in 2016 the sample-return mission OSIRIS-REx launched for asteroid Bennu, and in January 2017 NASA selected two more minor-body missions, to the Jupiter Trojans and to the metal world 16 Psyche. Humankind's understanding of the solar system has never been more comprehensive and, coupled with burgeoning results from extrasolar planet surveys, we are beginning to understand the fundamental rules governing planetary and solar system development in general. Of course, mysteries remain in our own backyard: we still do not know how many icy satellites orbiting the gas giants host subsurface oceans, prospective abodes for alien life; the history of when and why Mars lost its water is not yet fully understood; and the tantalizing possibility of an epoch of oceans on Venus is untested. But one major gap in our knowledge of our planetary system was filled by NASA's enormously successful MESSENGER mission to Mercury, which ended in 2015 after three fly-bys and four years of orbital operations and gave us our most comprehensive view of that world to date. This mission, whose full name was “MErcury Surface, Space ENvironment, GEochemistry, and Ranging”, showed us a planet wracked by impacts but still possessing an internal magnetic field, with a gigantic iron core and remarkably thin outer rock layer, a surprising geochemical make-up and an active early life followed by a protracted period of geological quiescence. Indeed, much of the planet's geological character directly reflects a simple evolutionary path: as Mercury aged and lost heat to space, its internal geological processes began to stutter before eventually all but stopping. Such a geological history arguably parallels most closely that of the Moon, but it is the compelling evidence that Mercury has visibly shrunk over the last four billion years or so that places it as an end-member of planetary behaviour among its solar system brethren. Early exploration Because of its size – halfway between the Moon and Mars – and proximity to the Sun, Mercury is a challenge to explore. Too deep in the star's gravity well to enable a spacecraft to slow down and achieve orbit easily, and too close to its glare to be safely observed by large telescopes, the smallest terrestrial planet remained unexplored until the arrival of NASA's Mariner 10 mission in the mid-1970s. Mariner 10 flew past Mercury three times in 1974 and 1975 but, because of a quirk in the heliocentric orbit of that spacecraft and Mercury's 3:2 spin-orbit resonance, only imaged about half of the planet's surface. Although literally half the planet remained hidden from us, images returned by this mission provided the foundation with which to understand Mercury's geological history, and its similarities and differences to its terrestrial planetary neighbours. What was immediately obvious with Mariner 10 data was the profusion of impact craters across the imaged hemisphere of Mercury (figure 1; Strom et al. 1975). This surface texture was redolent of the lunar highlands, ancient crust on the Moon that records billions of years of impact bombardment. Mariner 10 even hinted at a truly colossal impact basin: a glimpse of the eastern third of this feature, named Caloris, suggested a diameter of 1300 km. 1 View largeDownload slide Mariner 10 found Mercury's surface to be characterized by two dominant surface units: intercrater plains (centre) and smooth plains (right). Note the eastern rim of the Caloris basin visible at left. This image was taken near dusk, with the Sun shining from the right; the terminator is located near the centre. Shadows aid in the recognition of linear tectonic features, such as the wrinkle ridges that are oriented north–south here and that are manifest as narrow, sinuous arches. 1 View largeDownload slide Mariner 10 found Mercury's surface to be characterized by two dominant surface units: intercrater plains (centre) and smooth plains (right). Note the eastern rim of the Caloris basin visible at left. This image was taken near dusk, with the Sun shining from the right; the terminator is located near the centre. Shadows aid in the recognition of linear tectonic features, such as the wrinkle ridges that are oriented north–south here and that are manifest as narrow, sinuous arches. But impact craters were not uniformly distributed across the planet: some regions were notably devoid of large craters and basins, in contrast to other, heavily scarred areas. The relatively untouched areas were referred to as “smooth plains” and a view emerged that at least some such plains were in fact lava deposits (Strom et al. 1975, Trask & Guest 1975) – the reasoning being that lava is an effective resurfacing agent that could plausibly bury pre-existing craters and so render a surface younger-looking than it actually was. This view was not universally held (Wilhelms 1976). Comparison of surface material on the Moon thought to be lava but ultimately found to be impact ejecta when brought back to Earth by Apollo astronauts raised the prospect that similarly smooth areas on Mercury might also be the products of asteroid and comet strikes. This question remained open after the Mariner 10 mission. Yet that spacecraft did provide significant insight into the tectonic and thermal characteristics of Mercury. Small ridge-like landforms similar to examples observed on the Moon were reported, primarily within smooth plains deposits (figure 1). These “wrinkle ridges” were taken to primarily indicate subsidence of those deposits after emplacement (Murray et al. 1975). In addition, the identification of long, positive-relief landforms termed “lobate scarps” (Strom et al. 1975) suggested wholesale deformation of the planet's crust; although some argued for a volcanic origin, the recognition of craters apparently shortened by these scarps indicated that the crust there had been pushed together, in some ways similar to how mountain ranges on Earth are formed. Of course, most mountains on our own world reflect convergent tectonic plate motion, but there is no evidence for plate tectonics operating on Mercury. The widespread occurrence of these scarps across the face of Mercury seen by Mariner 10 instead suggested a global-scale process of contraction, whereby the planet had physically become smaller (Strom et al. 1975). This conclusion was supported by post-Mariner assessments of the likely thermal evolution of Mercury for given internal compositions and behaviours (e.g. Solomon 1977, 1978), although key questions remained, including whether these scarps were present across the rest of the planet, the full amount to which the planet had contracted, the effect of this process upon the planet's volcanic history, and when this global shrinking started and stopped (e.g. Solomon et al. 2008). A volcanic world The question of whether volcanism had taken place on Mercury was settled by MESSENGER. For example, the recognition of anomalously shallow impact basins strongly suggested that some type of fluid agent had served to fill in those basins (figure 2), with lava a natural culprit (Head et al. 2008) – a view consistent with that proposed for the smooth plains units seen with Mariner 10 data. Another type of landform was observed by MESSENGER that bolstered the case for volcanism: relict craters on impact basin floors, almost entirely buried with just their rims visible (inset figure 2b; Head et al. 2008). Termed “ghost craters”, these features must have formed after the basin, or they would not have survived; their state of burial requires that they formed before whatever fluid later filled them. Although the heat from large impacts can liquefy the surface, this impact melt lacks sufficient volume to almost fill a basin. Moreover, because impact melt is formed during the impact process, it could not possibly fill craters that formed much later. 2 View largeDownload slide The infilling of an impact basin by lavas. (a) The 67 km diameter Eastman crater, interpreted as unfilled because its central peak structure remains visible (purple arrow), as do its terraced walls (blue arrow). (b) The 240 km diameter Kunisada basin, the size of which implies that both a central peak ring and wall terraces ought to be visible, but are not. In fact, this basin appears almost full. Two ghost craters in the centre of the basin are shown in the inset, which strongly suggest a sequence of basin formation, cratering of the floor, and subsequent effusive volcanic activity. 2 View largeDownload slide The infilling of an impact basin by lavas. (a) The 67 km diameter Eastman crater, interpreted as unfilled because its central peak structure remains visible (purple arrow), as do its terraced walls (blue arrow). (b) The 240 km diameter Kunisada basin, the size of which implies that both a central peak ring and wall terraces ought to be visible, but are not. In fact, this basin appears almost full. Two ghost craters in the centre of the basin are shown in the inset, which strongly suggest a sequence of basin formation, cratering of the floor, and subsequent effusive volcanic activity. MESSENGER found yet more indications for volcanic activity on Mercury. In dozens of places, smooth plains deposits appeared to have overprinted pre-existing terrain, suggestive of vast lava flows that poured across the surface, burying whatever they encountered (e.g. Murchie et al. 2008). Without another candidate fluid that could plausibly have existed on the planet, it quickly became apparent that Mercury bore a record of (in places, sustained) resurfacing by effusive lavas (Head et al. 2008). The recognition that many large smooth plains deposits have a different spectral signature from that of the surrounding landscape (implying a difference in composition, as expected for lavas sourced elsewhere), all but affirmed that most such regions on Mercury are volcanic in nature (Denevi et al. 2009). Data returned from orbit about Mercury enabled the MESSENGER team to map the global geographic distributions of these major volcanic deposits (Denevi et al. 2013). They are found across the planet, but most volcanic plains are situated in the northern hemisphere (figure 3). And although numerous examples are of sufficient areal extent to rival large igneous provinces on Earth (e.g. Jerram & Widdowson 2005), none comes close to the size of the enormous Borealis Planitia. 3 View largeDownload slide The mapped global distribution of smooth plains on Mercury, after Denevi et al. (2013). The pink units are those for which a volcanic origin has been all but affirmed, and for which crater-based model ages exist. The blue units are the remaining smooth plains units yet to be so assessed; many are likely volcanic, although some smaller deposits may be ponded impact melt. Together, these plains constitute a little more than a quarter of the planet surface; note the hemispherical dichotomy in smooth plains distribution. The yellow dots denote sites of explosive volcanism, after Thomas et al. (2014). 3 View largeDownload slide The mapped global distribution of smooth plains on Mercury, after Denevi et al. (2013). The pink units are those for which a volcanic origin has been all but affirmed, and for which crater-based model ages exist. The blue units are the remaining smooth plains units yet to be so assessed; many are likely volcanic, although some smaller deposits may be ponded impact melt. Together, these plains constitute a little more than a quarter of the planet surface; note the hemispherical dichotomy in smooth plains distribution. The yellow dots denote sites of explosive volcanism, after Thomas et al. (2014). This expansive lava deposit occupies about 7% of the planet's surface (Head et al. 2011). No obvious morphological, spectral or compositional boundaries were identified within these plains that might suggest component subunits. Furthermore, there are no statistically significant variations in the areal density of impact craters – which was a surprise. Relative and absolute model ages for planetary surfaces are found by assessment of crater superposition, under the assumption that the older a given surface the longer it has had to accrue a record of impact cratering. That the craters on Borealis Planitia are relatively evenly distributed across the region suggests that it is about the same age everywhere – leading to the conclusion that this portion of Mercury was covered by lavas in one gigantic outpouring event about 3.7 billion years ago (Ostrach et al. 2015). Volcanic plains Other immense volcanic plains (if not as large as Borealis Planitia) include those within the mighty Caloris basin, originally spied by Mariner 10 and confirmed by MESSENGER to be almost 1600 km across (Murchie et al. 2008). These plains, named Caloris Planitia, are replete with tectonic structures that developed after those lavas solidified. The basin perimeter is ringed by more lava deposits, mixed with some melt from the impact that formed the basin itself (Denevi et al. 2013). And, although many smaller volcanic units are dotted across Mercury, most of the other large deposits lie inside older impact features, such as the lava infills within Tolstoj, Beethoven and Rembrandt basins (figure 3; Denevi et al. 2013). Model ages for some of these other major smooth plains units reveal an unexpected consistency: these deposits were all in place by about 3.5 billion years ago (Byrne et al. 2016). Early MESSENGER results returned similar ages for the Borealis and Caloris Planitiae, 3.7–3.8 billion years (e.g. Murchie et al. 2008, Head et al. 2011), itself a notable finding. But with continued analysis, this same approximate age was found for portions of the plains around the Caloris basin, those within the major southern-hemisphere impact basin Rembrandt, and the plains inside the Tolstoj and Beethoven basins (Fassett et al. 2009, Denevi et al. 2013, Ferrari et al. 2015). Many small smooth plains units could have been emplaced much later, but MESSENGER has shown us that most of the planet's effusive, widespread volcanic activity came to an end in the first quarter of Mercury's history. Effusive volcanism on Earth is often accompanied by explosive eruptions, and there is plenty of evidence for such activity on Mercury. For instance, irregularly shaped depressions were recognized from early MESSENGER fly-by data: on the basis of their shape, and a spectrally distinct “halo” of fine-grained material often encircling them, these depressions were interpreted as volcanic vents where pyroclastic (i.e. explosive) eruptions had taken place (Kerber et al. 2009). These explosive sites are distributed across Mercury, although few occur within smooth plains deposits, and most lie either within impact craters or along large tectonic structures (Jozwiak et al. 2018) (figure 3). In contrast to their effusive counterparts, explosive volcanic eruptions continued well into the latter half of Mercury's history, with some taking place perhaps as recently as 1 billion years ago (Thomas et al. 2014). One additional aspect of Mercury's volcanic character is of note: the origin that is suspected for much of the crust that is not classified as “smooth plains”. Usually termed “intercrater plains”, this landscape is far more cratered, and thus presumably older, than the smooth plains. But observations suggest that most, if not all, of this cratered surface was originally volcanic. For example, some smooth plains units are cratered to the point of having almost the same texture as intercrater plains (Whitten et al. 2014); in places, intercrater plains have similar spectral properties as some smoother regions, suggestive of a similar starting composition (Murchie et al. 2015); and only minor differences in areal crater density have been found between some intercrater and smooth plains deposits (Byrne et al. 2016). And, curiously, the oldest parts of Mercury's surface have been dated to about 4.1 billion years (Marchi et al. 2013), substantially younger than the assumed 4.5 billion year age of the planet. This “missing” oldest surface, and the propensity of lavas to pour over the landscape burying pre-existing terrain, suggests that Mercury's earliest crust was lost under major, global, sustained effusive volcanic activity. A tectonic world There was little doubt, following the Mariner 10 mission, that Mercury had experienced tectonic deformation. And it quickly became clear upon MESSENGER's arrival that this deformation had indeed occurred globally, with both lobate scarps and wrinkle ridges observed across the planet. The consensus remained that these structures reflected crustal shortening, but extensional landforms (indicative of the crust being pulled apart) were, with a few exceptions, conspicuous by their absence. Recording the distributions of tectonic landforms on Mercury was facilitated by global image and topographic maps compiled from MESSENGER observations. Like their counterparts on the Moon (Bryan 1973), wrinkle ridges on Mercury are typically broad, steep-sided but low-relief rises that are generally symmetrical in cross-section and straight to sinuous in plan view (figure 4a; Byrne et al. 2014). Generally a few hundred metres tall and up to a few tens of kilometres long, they are almost entirely restricted to smooth plains units. Wrinkle ridges are generally interpreted to be the manifestation of some combination of thrust faulting and folding (e.g. Mueller & Golombek 2004). 4 View largeDownload slide Examples of tectonic structures on Mercury. (a) A set of wrinkle ridges in Borealis Planitia; the arrows show where ridges have formed atop the buried rims of ghost craters. (b) A major equatorial lobate scarp, an example that features a prominent change in direction along its course. (c) Small graben within ghost craters, again situated in Borealis Planitia. Here, the purple arrows show the rims of those relict craters. 4 View largeDownload slide Examples of tectonic structures on Mercury. (a) A set of wrinkle ridges in Borealis Planitia; the arrows show where ridges have formed atop the buried rims of ghost craters. (b) A major equatorial lobate scarp, an example that features a prominent change in direction along its course. (c) Small graben within ghost craters, again situated in Borealis Planitia. Here, the purple arrows show the rims of those relict craters. Lobate scarps (figure 4b) are larger and more symmetrical counterparts to wrinkle ridges, with some examples a few thousand metres tall and hundreds of kilometres long (Byrne et al. 2014); the very largest scarps are Mercury's version of mountain ranges on Earth. The scarp itself is probably a folded portion of crust, situated atop a thrust fault that in places may extend several dozen kilometres into Mercury's lithosphere. Such structures are notable for a body the size of Mercury, and reflect considerable deformation of the crust over a protracted period of time. And, whereas wrinkle ridges are dominantly located in smooth plains units, lobate scarps generally deform intercrater plains. Although not nearly as widespread as shortening structures, evidence for crustal extension also exists on Mercury. Linear, steep-sided depressions were seen with Mariner 10 images, and found to occur extensively across Caloris Planitia when MESSENGER observed the entirety of those plains (Murchie et al. 2008). The depressions, interpreted to be down-dropped crustal blocks bounded on each side by normal faults and thus termed “graben”, were also identified in numerous other lava-filled impact basins across the planet (e.g. Prockter et al. 2010), as well as within Borealis Planitia (figure 4c; Watters et al. 2012). Note, however, that virtually no extensional structures have been found on Mercury outside the effusive lava deposits. Without plate tectonics, only a few mechanisms can plausibly drive crustal shortening and thus the formation of wrinkle ridges and lobate scarps. Their relatively small sizes, variety of orientations, and restriction to smooth plains units suggests that wrinkle ridges may have formed primarily in response to subsidence of those deposits as they settled into the basins in which they ponded, an interpretation supported by numerical modelling (e.g. Blair et al. 2013). Lobate scarps, on the other hand, are too large and too widely distributed to originate from subsidence; instead, a global-scale process must be responsible. Several mechanisms can plausibly drive crustal shortening at planetary scales, including convection of the interior, immense impacts, and tidal interactions with neighbouring bodies. However, each process leaves tell-tale signatures, none of which are observed for Mercury. Instead, the planet's population of lobate scarps attests to a history of planetary volume loss arising from the cooling, and contraction, of the core and mantle through geological time (Solomon 1978). Like wrinkle ridges, Mercury's extensional structures are likely to have formed from a lava-specific process. In addition to subsidence after ponding, lavas contract as they cool. But where a planet loses volume in three dimensions, lavas contract horizontally – placing the upper surface of flows (or packets of flows) into extension. This thermal contraction drives the development of structures that eventually grow to the graben MESSENGER observed in a manner analogous, say, to how the Giant's Causeway in Northern Ireland formed (Freed et al. 2012). The one exception to this mechanism may be a radial set of extensional structures within the Caloris basin, which are termed Pantheon Fossae. The origin of these graben remains an open question, although some combination of thermal contraction and magmatic intrusion might be involved. A full assessment of Mercury's tectonic character necessarily includes estimates of when different styles of tectonic deformation, and thus their formative mechanisms, operated on the planet. In contrast to lava plains, however, the utility of areal crater densities in determining the ages of tectonic landforms is limited. This is because, where the emplacement of a lava flow occurs as a single event with a single age, tectonic structures can remain dormant for some time before being reactivated if they are favourably aligned with a prevailing stress field. Thus the age of a ridge, scarp or graben as derived from crater statistics may correspond only to the latest tectonic movement in a series of such events, and so the full tectonic history of an area, or the length of time in which a given process operated, is difficult if not impossible to determine. Nonetheless, superposition relationships offer some insight into the timing of tectonic deformation. For example, that wrinkle ridges occur within ponded lava deposits indicates that the ridges formed after those lavas were emplaced: if it were the other way around, the ridges would show evidence of being partially buried by those flows, and such observations have not been made for Mercury. Indeed, nowhere have lavas been seen to embay ridges or scarps, suggesting that both lava subsidence and global contraction began to assert themselves after the bulk of the planet's effusive lava plains were emplaced (Strom et al. 1975, Solomon et al. 2008, Banks et al. 2015). The same argument holds for Mercury's graben, almost all of which are situated within smooth plains deposits. Mercury's scarps may have started to grow during the last episode of effusive volcanic activity on the planet, but whenever they began to form they certainly persisted beyond the end of this episode. This observation, and the finding that most major lava plains were in place by about the same time, strongly suggests that the histories of volcanism and tectonism on Mercury are intertwined. And the planetary-scale process invoked to account for the planet's scarps – global contraction – offers an explanation for why this might be so. Global contraction and volcanism Simply put, the interior of a planet starts off very hot and then begins to cool. It will do so in a number of ways: by convection of a molten or fluid interior portion, conduction through a solid portion, and/or radiation from that solid portion (often the upper part of the body) to space. A direct, geometric consequence of this process is the loss of its internal volume as it cools; if its outer portion is solid then the volumetric loss could be recorded there by tectonic landforms (e.g. Solomon 1977). Several interior processes may serve to enhance or inhibit this process; for instance, the freezing of a solid inner iron-rich core can augment the rate at which a body contracts, whereas the decay of radiogenic elements in the mantle can provide additional heat to offset interior cooling. The possibility of Mercury having shrunk as it cooled was raised from Mariner 10 observations (Strom et al. 1975), and served as the basis for studies after that mission ended (e.g. Solomon 1977, Hauck et al. 2004). Mariner 10-based estimates of how much the planet had contracted – a key variable for thermal evolution models, which seek to characterize not only how the planet has behaved through time but also its probable chemical and compositional stratigraphy – remained at odds with model predictions (Hauck et al. 2004), until MESSENGER observations resolved the issue (Byrne et al. 2014). Now there is a consensus that the planet has contracted and that this phenomenon had an enduring effect on the planet's geology. Perhaps most obviously, global contraction leads to tectonic deformation. A decrease in volume of a planetary body translates to horizontal compression of the upper surface (Solomon 1978). Such a stress state drives portions of the crust up and over each other, and this deformation is accommodated by the formation of large folds atop thrust faults – Mercury's lobate scarps. Moreover, this process ought to operate approximately evenly across the surface, leading to structures in a variety of orientations. Detailed mapping of Mercury's scarps shows us that, by and large, these landforms show no preferred alignments across the planet (Byrne et al. 2014). There is another geological consequence of global contraction. A cooling interior will be less likely to generate magma that ascends and ultimately erupts as lava, but the stress state that leads to crustal shortening also hinders magma ascent in the first place (figure 5; Solomon 1977). Under a scenario where stresses are approximately equal in the horizontal and vertical direction, magma has an equal chance of spreading vertically or horizontally at depth, and of reaching the surface and erupting (figure 5a). A horizontally extensional regime favours magma reaching the surface, and this is further enhanced if the melt happens to be rich in volatiles such as water or sulphur dioxide, which would increase the likelihood of an explosive eruption (figure 5b). But global contraction applies a stress regime to the near-surface region that stops magma from ascending right to the surface, driving it into horizontally extensive intrusions called sills and laccoliths, instead of permitting it to erupt (figure 5c). 5 View largeDownload slide Major tectonic stress regimes. (a) A neutral regime, under which magma is equally free to ascend or migrate horizontally. (b) An extensional regime, under which magma is more likely to reach the surface. (c) A contractional regime, the most inimical to successful magma ascent. Here, σv is the vertical stress and σH and σh are the maximum and minimum horizontal stresses, respectively. Pm is magma pressure. 5 View largeDownload slide Major tectonic stress regimes. (a) A neutral regime, under which magma is equally free to ascend or migrate horizontally. (b) An extensional regime, under which magma is more likely to reach the surface. (c) A contractional regime, the most inimical to successful magma ascent. Here, σv is the vertical stress and σH and σh are the maximum and minimum horizontal stresses, respectively. Pm is magma pressure. Global contraction is therefore an effective mechanism by which a planet can develop a planet-wide distribution of large thrust faults and folds, and experience a reduction and ultimately a cessation of volcanic activity – exactly as is observed on Mercury. And, by this reasoning, the model ages of the planet's major volcanic smooth plains tells us when Mercury's global contraction had become fully manifest: by around 3.5 billion years ago (Byrne et al. 2016). Of course, there is plenty of evidence for explosive volcanism having long outlived its eruptive counterpart, perhaps by as much as a couple of billion years (Thomas et al. 2014). However, even this observation is consistent with a sustained record of global contraction for Mercury: almost all sites of explosive activity are collocated with major weaknesses in the crust, such as within and around impact craters and basins, and along deep-seated scarp faults (Jozwiak et al. 2018). For any magma to erupt against a prevailing contractional stress regime, it would be best placed to do so were it both volatile-rich and able to utilize deep faults along which it could rise (Hamilton 1995). Conclusion The MESSENGER mission has revolutionized our understanding of Mercury, providing a global basis for determining how, when, and why its geological forces have made it look the way it does. There is a remarkable linkage between the planet's volcanic and tectonic histories, perhaps more so than on any other rocky solar system body. Although a global population of lobate scarps is present on the Moon, and most effusive volcanism ended on that body billions of years ago, lunar scarps are far smaller than those on Mercury, and extensional structures are more common on the Moon. Mars, too, hosts lobate scarps, but also features a protracted history of effusive volcanism that resurfaced almost half of the planet, in places as recently as a few hundred million years ago. And the tectonic and volcanic histories of Venus and Earth are yet more complex, with plate tectonics dominating the geodynamics of our own world. Mercury may therefore represent an end-member for how small, rocky planets typically behave. If so, then by fully characterizing that body's geological character, we may be better placed to understand the geology of extrasolar planets. With the recent detection of a Mercury-sized exoplanet (Barclay et al. 2013), and future missions focused on discovering yet more Earth-like exoplanets, (including the James Webb space telescope and the Transiting Exoplanet Survey Satellite), we are developing the tools to understand how rocky planets develop and evolve in general. But we are not yet finished with Mercury. In its elliptical orbit about the planet, MESSENGER came closest to, and so returned the best data for, the northern hemisphere only. Open questions therefore remain, including the interior structure of the planet's southern hemisphere, the geochemical properties of rocks on the surface at scales below what MESSENGER was able to resolve, and even the long-term behaviour of the planet's magnetic field. More broadly, we still do not know when global contraction definitively started, nor do we fully understand how Mercury's intercrater plains came to be, or even what lies beneath that heavily cratered landscape. Fortunately, the ESA BepiColombo mission (Benkhoff et al. 2010), due to launch in late 2018, will be able to address at least some of these questions, and with luck even later missions will follow, to land, rove and even return samples to Earth one day. Until then, we must content ourselves with the data to hand and our nascent but growing understanding of the geological history of the innermost planet. ACKNOWLEDGMENTs My profound thanks to the entire MESSENGER team, who successfully designed, dispatched and navigated a desk-sized but incredibly capable spacecraft to a strange world more than 70 million km away. Thanks too to numerous co-authors who helped shape the concepts described here, including but not limited to Christian Klimczak, Sean Solomon, Celâl Sengör, Steven Hauck, Brett Denevi, Maria Banks, Clark Chapman, Alexander Evans, Caleb Fassett, James Head, Francis McCubbin, Lillian Ostrach, Thomas Watters and Jennifer Whitten. Finally, thank you to A&G for the opportunity to fly the flag for MESSENGER and Mercury, an opportunity I'll always gladly take because that planet is so weird. REFERENCES Banks M Eet al. 2015 J. Geophys. Res. Planets 120 1751 CrossRef Search ADS Barclay Tet al. 2013 Nature 494 452 CrossRef Search ADS PubMed Benkhoff Jet al. 2010 Planet. Space Sci. 58 2 CrossRef Search ADS Blair D Met al. 2013 J. Geophys. Res. Planets 118 47 CrossRef Search ADS Bryan W B 1973 Proc. Lunar Sci. Conf. 4 93 Byrne P Ket al. 2014 Nature Geosci. 7 301 CrossRef Search ADS Byrne P Ket al. 2016 Geophys. Res. Lett. 43 7408 CrossRef Search ADS Denevi B Wet al. 2009 Science 324 613 PubMed Denevi B Wet al. 2013 J. Geophys. Res. Planets 118 891 CrossRef Search ADS Fassett C Iet al. 2009 Earth Planet. Sci. Lett. 285 297 CrossRef Search ADS Ferrari Set al. 2015 Geol. Soc. London Spec. Publ. 401 159 CrossRef Search ADS Freed A Met al. 2012 J. Geophys. Res. 117 E00L06 Hamilton W B 1995 Geol. Soc. London Spec. Publ. 81 3 CrossRef Search ADS Hauck S Aet al. 2004 Earth Planet. Sci. Lett. 222 713 CrossRef Search ADS Head J Wet al. 2008 Science 321 69 CrossRef Search ADS PubMed Head J Wet al. 2011 Science 333 1853 CrossRef Search ADS PubMed Jerram D A & Widdowson M 2005 Lithos 79 385 CrossRef Search ADS Jozwiak Let al. 2018 Icarus 302 191 CrossRef Search ADS Kerber Let al. 2009 Earth Planet. Sci. Lett. 285 263 CrossRef Search ADS Marchi Set al. 2013 Nature 499 59 CrossRef Search ADS PubMed Mueller K & Golombek M 2004 Annu. Rev. Earth Planet. Sci. 32 435 CrossRef Search ADS Murchie S Let al. 2008 Science 321 73 CrossRef Search ADS PubMed Murchie S Let al. 2015 Icarus 254 287 CrossRef Search ADS Murray B Cet al. 1975 J. Geophys. Res. 80 2508 CrossRef Search ADS Ostrach L Ret al. 2015 Icarus 250 602 CrossRef Search ADS Prockter L Met al. 2010 Science 329 668 CrossRef Search ADS PubMed Solomon S C 1977 Phys. Earth Planet. Inter. 15 135 CrossRef Search ADS Solomon S C 1978 Geophys. Res. Lett. 5 461 CrossRef Search ADS Solomon S Cet al. 2008 Science 321 59 CrossRef Search ADS PubMed Strom R Get al. 1975 J. Geophys. Res. 80 2478 CrossRef Search ADS Thomas R Jet al. 2014 Geophys. Res. Lett. 41 6084 CrossRef Search ADS Trask N J & Guest J E 1975 J. Geophys. Res. 80 2461 CrossRef Search ADS Watters T Ret al. 2012 Geology 40 1123 CrossRef Search ADS Whitten J Let al. 2014 Icarus 241 97 CrossRef Search ADS Wilhelms D E 1976 Icarus 28 551 CrossRef Search ADS © 2018 Royal Astronomical Society
Astronomy & Geophysics – Oxford University Press
Published: Feb 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.
All for just $49/month
Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.
Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.
It’s easy to organize your research with our built-in tools.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera