Water from Heaven

Are we alone in the universe? Today, our planet Earth, with atmosphere, clouds, and oceans, surrounding a mostly solid mass made of heavy elements, appears to be the sole abode of life among the planets orbiting our Sun. On dry and frigid Mars, we can discern some vestiges of a gentler climate, of a liquid age when water flowed on the planet’s surface. A few peculiar meteorites, fragments from Mars found on Earth, harbor features that some researchers have identified as fossil Martian bacteria. Others disagree, and the question is not settled.¹ Could some such microbes still survive in hibernation somewhere in the long frozen soil of the red planet, awaiting the return of better days? Could human colonists, engaging in an activity of “terraforming,”² render the Martian climate less hostile to life? All these questions and more are raised as we terrestrials wonder how far and how fast to push the exploration that began when the Viking-1 and -2 spacecraft reached Mars in 1976. Increasingly sophisticated robot equipment, with no requirement for a round-trip ticket, has sent us a wealth of data in recent years, answering some questions and raising new ones.³

And elsewhere? Dry Venus, with its thick atmosphere (more like a gaseous ocean) of carbon dioxide, has temperatures at the surface that are infernal, and the planet hardly seems hospitable to life. Things are no better on Mercury, with no atmosphere and temperatures oscillating regularly between infernal heat and deadly cold. And the giant planets are but failed stars, no place for life anything like what we know. What about their satellites? Europa, a moon of Jupiter, and Titan, a moon of Saturn, appear to be covered with ice, perhaps floating on oceans despite the extreme cold.⁴ Fascinating questions, but we might as well wait for the data that the Huygens-Cassini space probe will send back.

Ever since Copernicus put the Sun, rather than the Earth, in its central position, we call our planetary system a solar system. Is it unique? Along with the Copernican Revolution came a realization of the vastness of the universe, a realization that for stars to be visible at all at their enormous distances, they must be suns; and by the same token, the Sun is just an ordinary star. Once this is accepted, it is natural to ask whether there could not be many planetary systems and indeed inhabited planets, around other stars in the universe.⁵ For our world of earth and water to exist, stars had to be born and die, and our Sun and its retinue of planets be formed from the debris of past generations of stars.

A few thousand stars are visible to the naked eye, and astronomers count thousands of billions of them. But, at best, only a dozen or so planetary systems have been conclusively identified so far. It is nevertheless believed that planetary systems abound in the universe, that the formation of planets often goes along with the process of star formation. In order for a cloud of matter to become a star, it has to contract, becoming denser and hotter. However, while the mutual gravitational attraction of the matter in the cloud tends to pull it together, the cloud’s rotation will accelerate at the same time. This results from the law of conservation of angular momentum: anything that spins continues to spin unless its spin is slowed down by increasing another body’s spin, and the “angular momentum” that measures the spin depends on the spin rate and on the distance from the axis of rotation. Just as a skater will spin more quickly by bringing her arms in toward her body, so the contracting cloud will rotate faster and faster. For the same reason, a comet in an extremely elongated elliptical orbit around the Sun moves much more quickly when it is close to perihelion than when it out beyond Jupiter’s orbit. Everything in the universe is rotating: galaxies, nebulae, stars, planets and their satellites. It is hard to believe that all stars are born from the very few clouds that are not rotating. However, as a contracting cloud of gas and dust spins more and more quickly, it is distorted by centrifugal force and ends up losing mass in its equatorial plane in a sort of pre-stellar wind. Can it ever contract enough for the temperature and density in its central core to reach the “ignition point” for thermonuclear reactions, to start it off on the life of a true star?

Despite these difficulties, stars shine. How do they rid themselves of their angular momentum? For many of them, the answer is that the cloud has split up to form a binary or multiple stellar system: the angular momentum is still there, but it’s been concentrated in the motion of the two or more stars around each other. Some single stars—always very young—spin very quickly on their axes, but many single stars rotate very slowly (in 25 to 27 days in the case of our Sun). However, although the angular momentum of the Sun itself is quite small, that of the solar system as a whole is not. Although the planets contain only 0.13 percent of the mass of the Sun, their orbital motions account for nearly all the angular momentum of the solar system, 15,000 times more than the rotation of the Sun alone on its axis. In the process of formation of the Sun and its retinue of small and giant planets, nearly all the angular momentum was somehow transferred from the central condensation or “proto-sun” to the surrounding “protoplanetary” cloud. This transfer process depends on the existence of magnetic lines of force that accelerate the distant cloud’s rotation at the expense of the spin of the central condensation. Not all the details have been worked out, but it’s hard to see an alternative to this scenario. Within the flattened spinning cloud, protoplanetary condensations must have formed, becoming the planets and their satellites quite quickly (in less than 100 million years). All the planets revolve around the Sun in the same direction and in nearly the same plane (the ecliptic) close to the equatorial plane of the rotation of the Sun itself, and nearly all the planets rotate on their own axes; and for those that have satellites, those satellites also revolve around them in the same direction. Such a scenario is not likely to be peculiar to the solar system; it accounts for the multitude of apparently single slowly spinning stars. And while only a few planets have been identified with certainty outside the solar system, astronomers have shown, mostly using infrared observations, that many stars are surrounded by disks of gas and dust. Far from being an exception, planetary systems may be quite common.

Interstellar, protostellar, and proto-planetary-system clouds contain hydrogen, helium, and other elements in gaseous form as atoms, ions, and molecules. The H₂O molecule must be there. These clouds also contain solid grains of different sorts, sometimes coated with sheathes of ices, often water ice. As gravitational contraction proceeds, other grains may condense and trap more water. However, in the gestation that preceded the birth of our Sun, the gradual warming of the protosolar cloud leading to the firing up of the thermonuclear furnace also warmed up the protoplanetary cloud. Far from the center, temperatures remained low, so that even grains composed of volatile substances such as water, methane, and ammonia survived in solid form. But closer to the central source of warmth, as temperatures rose above 1,000 or even 1,500 degrees K (about 2,400°F), water ice evaporated, and only more or less refractory grains survived together with silicates and metallic iron. In between, other minerals could be found, in particular hydrated minerals in which water molecules (in fact broken up into hydrogen atoms and OH radicals) were trapped in the mineral crystal structure. At this point, a strong “wind” from the young Sun swept through the flattened cloud, in its central part sweeping away nearly everything not solid (or not attached to a solid grain). In the inner part of the solar system, where today orbit Mercury, Venus, the Earth-Moon couple, and Mars, nearly all of the mass, practically all the hydrogen and helium, was swept away and lost for the formation of these planets. The helium found on Earth, extremely valuable for safely flying balloons and dirigibles, is not a residue of the Big Bang but rather a product of the radioactive decay of uranium formed in earlier generations of stars. The argon-40 isotope, nearly 1 percent of air, comes from the decay of radioactive potassium. The other argon isotopes, like the other rare gases (neon, krypton, xenon) are extremely rare (both in an absolute sense and compared with their abundance in the Sun), as a result of the violent protosolar wind that cleaned out the inner protoplanetary clouds. These inert gases could not form compounds with other elements or be bound up in minerals.

Farther out, between the orbits of Mars and Jupiter, hydrogen and helium were also swept away. Here, source of most meteorites, orbit the asteroids, possibly leftovers of a shattered planet or maybe of one that never formed. Farther away still, the protosolar wind was weaker, and the protoplanetary condensations that later became Jupiter, Saturn, Uranus, and Neptune kept most of their gaseous hydrogen and helium; indeed their composition is pretty close to that of the Sun. So this, painted in broad strokes, is how our solar system was formed. There are of course plenty of features that still need to be filled in, as astronomers try to explain in detail the differences in chemical and isotopic composition among the different planets, comets, and meteorites.

Buried in the Earth’s crust, miners can extract—blessing or curse?—uranium and other radioactive elements made in a supernova explosion less than five billion years ago. Old Mother Earth is still young enough to have kept some traces of this event. After ten billion years, the uranium would be much rarer, half of it decaying to lead every 4.5 billion years. Whether from meteorites fallen to Earth, or rocks from the Moon or Mars, the isotope ratios all point to condensation of the solid bodies of the solar system between four and five billion years ago. Which ratios? Generally, the ratio of the abundance of a radioactive isotope to that of the stable isotope resulting from its decay (for example, the ratios uranium-238/lead-206, uranium-236/lead-207, potassium-40/argon-40, etc.). Earth’s childhood having been particularly tumultuous, few rocks older than three billion years are to be found today. Nevertheless, the specialists are quite convinced that, give or take a few hundred million years, the solar system was formed 4.6 billion years ago.

At the very beginning of this process, the condensation destined to become Earth was surrounded by a cloud consisting mainly of hydrogen and helium, with some water vapor as well. However, this primary atmosphere of the Earth must have disappeared in the course of the clean sweep by the young Sun’s radiation and wind. But where then does the water of the Earth come from? Did heavenly waters arrive on the surface of an already formed planet? For a billion years following the birth of the Earth, its surface, like the surfaces of the Moon, Mars, and Mercury, was intensely bombarded by meteorites, some of which contained water.⁶ The marks of these impacts have, for the most part, long been erased by erosion on Earth, to a lesser extent also on Mars, but they are still clear on the pockmarked surfaces of the Moon and Mercury. There are over a billion billion tons of water on Earth, and it is difficult to calculate exactly how many tons of impacting meteorites were necessary to supply it. The water needed to fill the oceans must have come with enough debris to make a layer a hundred kilometers thick, much thicker than the Earth’s crust as we know it today.

Comets may be another source. Far beyond the orbits of the giant planets, small clumps of matter must exist very nearly undisturbed since the beginning: a little dust, surrounded by water and other ices. These “dirty snowballs,” as Harvard astronomer Fred Whipple called them, only reveal themselves as comets after perturbation of their motions by a passing star, sending them into extremely elongated elliptical orbits that take them into the inner part of the solar system. Closer and closer to the Sun, warmer and warmer, some of the ices evaporate, and a cloud of gas and dust escapes from the semisolid nucleus of the comet. Swept away from the Sun by the pressure of its radiation and of its wind of particles, the cloud becomes a coma (hair) and sometimes a splendid tail shining in the skies for several weeks.⁷ Today, fortunately, major comet impacts on the Earth are rare—at least we’d like to think so. And earlier?

Are catastrophic comet collisions needed to carry water to Earth? There has been a long-running scientific controversy about the possibility that “micro-comets” might constitute a continuing supply of water for the Earth, even today. In 1981, Iowa University physicist Louis Frank was studying measurements of ultraviolet radiation from the night side of the Earth using an instrument on board the NASA Dynamics Explorer 1 satellite, and he identified spectral emission lines due to upper atmosphere nitrogen and oxygen atoms, excited by the particles of the solar wind.⁸ On the images, Frank found many black spots, which he attributed to comets of very low mass (10 to 30 tons). Frank’s estimate was that twenty or so such micro-comets collided with the Earth’s atmosphere every minute, bringing enough water to cover the Earth with several centimeters (or a few inches) extra every 20,000 years. If such a bombardment went on for three billion years, it would be enough to renew all the water in the oceans many times over. Following the launch in 1996 of NASA’s Polar satellite, Frank also found ultraviolet emission lines due to the hydroxyl (OH) radical, and he argued that this confirms the idea that these objects contain water. However, no signs of such frequent impacts have been observed on the Moon or Mars. Neither are there signs of water enrichment in the atmospheres of the giant planets, even though they are much more powerful collectors of interplanetary debris than the Earth. Could the lines identified with OH, emitted at over a hundred kilometers altitude, be due to water molecules already present in the Earth’s atmosphere? That too is hard to believe: the upper atmosphere is extremely dry, and water vapor seldom gets above the level of the tropopause (10–15 km altitude). Many specialists who have looked at the data believe that the black spots are instrumental artifacts and not due to atmospheric features, but Louis Frank and his colleagues have not conceded defeat. The puzzle remains.⁹

Water arrived from the skies, but the atmospheric water of the first ages of our planet must have been lost if not recycled in the Earth long ago. Water arrived as the first solid matter condensed to form the proto-Earth, it arrived during the accretion phase as the planet acquired its final mass, perhaps also during the following billion years. It arrived in solid or frozen granules and in the hydrated minerals that accumulated as the planet formed, perhaps also with the rain of comets and meteorites, the debris of asteroids. These granules are cold as they arrive, but the impact shock of agglomeration and accretion heats up the surface. At the same time, the (proto-)planetary surface absorbs solar radiation, converting it into heat, a process that necessarily continues today. Moreover, the process of consolidation of the infant Earth took place only a short time (maybe 300 million years) after the cloud contracting to form the solar system was enriched by a supernova explosion. The intense radioactivity of short-lived nuclei produced by such an explosion must have heated each of the proto-planets still more, not only at its surface but throughout its volume. Even today, over four and a half billion years later, the long-lived radioactive nuclei produced by the same supernova are responsible for much of the heat flux emerging from inside the Earth: about a tenth of a watt per square meter (that still adds up to 50 million megawatts for the whole planet), very little compared with the 240 watts per square meter absorbed from sunlight (120 billion megawatts for the planet). Most geophysicists believe that water, like the other components of the ancient secondary (not primary) atmosphere, emerged as a result of the “outgassing” of the outer solid layers of the Earth.

Today, the oceans cover 71 percent of the planet’s surface; their average depth is 3,700 meters (over 12,000 ft.), so that they contain 1.35 billion cubic kilometers or 1,350 billion billion liters of water. Enormous as such a figure is, it still represents only 0.02 percent of the total mass of the Earth. The only way to get such a huge quantity out of the Earth’s interior was to make it boil. Even though the proto-Earth started to form in the cold, Earth’s early infancy took place in heat released by intense radioactivity over the following few hundred million years. Much of the abundant metallic iron of the first condensations must have melted, heated by the accretion of still more iron-rich material, and this molten iron must have flowed down toward the center of the Earth, forming a solid metallic inner core surrounded by liquid. Together, this iron and nickel core contains about a third of the mass of the Earth, with temperatures reaching 7,000 °C (12,632°F). The slow infiltration of liquid iron to the central region is in effect a particularly gentle sort of fall, converting gravitational energy into heat just as in the case of the violent fall of meteorites. The intense heat released inside the Earth at its formation, reinforced and trapped by the inner core’s solidification, supplemented by the heat of radioactive decay, can escape only through the mantle, intermediate layers of less dense material that comprise two-thirds of the Earth’s mass. Although not liquid, at temperatures of a few thousands of degrees Kelvin, the mantle is plastic (in the technical sense), i.e., it can flow, and the convective flow transfers heat outward. The temperature structure of the mantle and the strength of its convection depend on the temperature in the central core, the radioactive release of heat in the mantle, and the rate at which heat can be lost at the upper surface.

Under these conditions, certain minerals and elements tend to separate from others, ending up near the surface of the globe in a thin chemically different crust (up to 70 km or 43 mi. thick, only 0.4% of the mass of the Earth). This is the lithosphere, less dense but much more rigid than the mantle. Below it, in the mantle, heated minerals give up water, carbon dioxide (CO₂), and other gases that they contain, and these gases tend to rise. Here and there they find their way to the surface through cracks and vents in the rigid lithosphere, emerging from the smoking mouths of volcanoes. Such a process, no doubt many times more intense than today, must have given rise to the planet’s “secondary” atmosphere.

The slow settling of iron, together with radioactivity, have kept the center of the Earth hot, even while temperatures at the surface have stayed in a reasonable range over most of the Earth’s history. And yet the Sun was almost certainly fainter in the past, perhaps 40 percent fainter some four billion years ago. If the Earth’s atmosphere then was the same as today’s, the Earth would have been completely frozen over.¹⁰ But today’s atmosphere, rich in oxygen but relatively poor in carbon dioxide, is a product of the evolution of land plants. At first, the atmosphere contained very little oxygen; on the contrary, it contained much more CO₂ and other gases. These gases contribute to what is known as the greenhouse effect, trapping thermal infrared radiation (invisible heat radiation, with wavelengths greater than four microns) in the lower layers of the atmosphere. By contrast, solar visible and near infrared radiation at shorter wavelengths passes quite easily through the atmosphere. The greenhouse gases, molecules made of three atoms or more (H₂O, CO₂, NH₃, CH₄, etc.), are relatively rare in the atmosphere, but they are effective absorbers of infrared radiation. On the contrary, the nitrogen (N₂) and oxygen (O₂) molecules have structures far too simple for them to interfere very much either with sunlight or with the thermal infrared radiation emitted by the Earth. Three billion years ago, the natural atmospheric greenhouse effect must have been much stronger than today, keeping the Earth’s surface warm despite the fainter Sun of those days.

As outgassing of the Earth’s mantle proceeds, spouting more and more water vapor into the atmosphere, the greenhouse effect increasingly warms the surface and lower atmosphere. However, when the relative humidity reaches 100 percent—in other words, when the air is saturated with water vapor—the water must condense. Clouds of water and ice form, everywhere, and one can imagine the deluge. So long as volcanoes continue to spit out enormous volumes of water vapor, the air stays saturated, clouds pile up and burst, and water falls from the sky to form the oceans. The process here painted with the broadest of brush strokes certainly lasted much longer than forty days and forty nights. Even with 10 cm (4 in.) of rain per day, every day, everywhere, almost a hundred years would be needed to fill the oceans, provided there was no evaporation. And even if this deluge took a thousand or a million times longer, it would still be but a brief moment in the history of the Earth, a moment in which everything was forever changed. The geological record of sedimentary rocks, formed in the presence of liquid water, proves that oceans have existed for at least three billion years, maybe even four, and that their volume has hardly changed for the last two billion. Outgassing of the interior continues, but at a much reduced rate, most of the volatile material having already been lost to the surface and atmosphere. And the gases that emerge from volcanoes today result to a large extent from recycling between oceans and mantle.

What happened elsewhere? It appears that the Earth was just in the right place to keep its oceans.¹¹ Further away from the Sun and therefore receiving less sunlight than the Earth, Mars was colder. And with lower mass and density and a weaker magnetic field, the red planet had less internal (“areological” rather than geological) activity than the Earth; whatever gases were emitted by its now extinct volcanoes, they tended to condense frozen on the ground, so that having a much weaker greenhouse effect than the Earth, Mars remained cold. On Venus, closer to the Sun and very similar to Earth in size so that volcanic activity may have been intense, a thick atmosphere developed under much warmer conditions. Although the atmosphere must have contained water vapor initially, condensation of water was rare, and in the long run the water vapor was lost by the process of photolysis, solar radiation breaking apart the H₂O molecules. The hydrogen thus freed easily escaped the gravity of Venus. However, while on Earth most carbon dioxide has been locked up in the crust in the form of carbonates by way of chemical reactions that depend on the presence of liquid water, on Venus, without any ocean, this was not possible. On the contrary, the carbon dioxide accumulated as a gas in the atmosphere, leading to a runaway greenhouse effect, higher and higher temperatures, more and more CO₂ leaving the surface to join the atmosphere, reinforcing still more the greenhouse effect. For Venus, any chance of accumulating an ocean was lost.

We humans are landlubbers, living on continents and islands, land surfaces emerging more or less from the oceans. The 1,350 million cu. km of water could cover the entire globe with a single world ocean more than 2,500 meters deep; but in fact this mass of water is collected and partly separated in the seven seas, with average depth 3,730 meters. The biblical “All the rivers flow into the sea, and the sea is not filled; and the rivers continue to flow” in fact raises at least two questions. What we generally call the water cycle—evaporation, precipitation, runoff—does indeed explain that the rivers continue to flow, and that the sea does not overflow. But the rivers also carry with them the debris of the land: dissolved salts and minerals, 30 billion tons of mud and other suspended matter every year. How can it be that the bottom of the sea has not filled up? The facts are there: the sea floor is not a many-mile-thick layer of silt and other debris accumulated in the course of the billions of years since the origin of the Earth and its oceans. How can it be that the continents and mountains are still standing and not eroded and washed away by falling rain and running water? Not all these questions have been settled (are they ever, when scientific research keeps raising new ones?), but since the 1960s all these questions have appeared under a new light following the great revolution in the Earth sciences and the establishment of the discipline of plate tectonics.

In 1912 the German meteorologist Alfred Wegener (1880–1930) formulated the hypothesis of continental drift, according to which the continents move relative to one another, separating or colliding. Two hundred million years ago, Africa and South America were joined in one supercontinent, Pangaea—a long time ago, but still quite recent in terms of the age of the Earth. Using lasers on the ground to illuminate reflectors on board artificial satellites, space-age surveyors can make very precise determinations of the positions of those lasers, and the results confirm that Europe and North America are drifting apart 3 cm (a little over an inch) every year. The sea floor, always less than 200 million years old, is extruded from the mantle along a colossal underwater mountain range—the mid-ocean ridge—and spreads out in both directions until it finally plunges into the depths of the mantle of the Earth in the subduction trenches (fig. 2.1).

Maps usually color the continents differently from the oceans. This differentiation of the surface of the Earth—separation into oceanic and continental domains—remains valid even apart from the presence of the liquid water. The material of the continents was chemically separated early in Earth’s history from the material of the sea floor. The not very dense continental crust is rich in silicates, but oceanic crust consists of basalt richer in minerals of iron and magnesium. The continental plates float above denser layers, like a sort of “scum,” as distinguished French geo-chemist Claude Allègre put it.¹² Both the composition and the structure exhibit basic differences. It is striking that on modern maps showing the topography of the sea floor as well as that of the continents, the entire ocean realm is organized around an underwater mountain range much longer than continental cordilleras (over 60,000 km—i.e., one and a half times the circumference of the Earth). Starting from the Arctic Ocean north of Spitsbergen, the mid-ocean ridge emerges from the ocean in Iceland. It then dives into the depths, and from north to south it divides the Atlantic into two roughly equal parts before turning east and passing south of Africa, the Indian Ocean, and Australia, finally extending north toward California and even Vancouver in British Columbia. A branch splits off in the Indian Ocean toward the Red Sea, while two others in the Pacific extend in the direction of Chile and the Galapagos Islands. The ridge, rising some 2,500 meters above the abyssal plain, with the surface of the sea about 2,500 meters above its summit, was hardly known before the 1950s. It differs from continental mountain ranges in its length, its composition, and its very peculiar structure: the crest of the ridge has a breadth of a few thousand kilometers, but it is split all along its length by a rift—a narrow valley (a few dozen kilometers), very deep in places, marked by numerous volcanoes and hot-water springs.

FIGURE 2.1 Sea-floor spreading and continental drift. At the top, a map of the tropical zone. From west to east: the eastern Pacific (Nazca plate), South America (Atacama subduction trench and the Andes), the Atlantic with the mid-Atlantic ridge separating the South American and African plates, and Africa. Dots indicate the epicenters of earthquakes. At the bottom, a vertical cross-section from the Nazca plate to Africa, with convective motions of the upper mantle indicated.

At present, and for the last several hundred million years at least, convection currents rise in the Earth’s mantle (fig. 2.1) and culminate under the ridge, bringing up heat along with water trapped in the minerals of the magma, activating undersea volcanoes and hot springs. About 20 cu.km (several billion tons) of basaltic rock emerge around the rift and mid-ocean ridge each year. In about 200 million years, that amounts to enough to renew completely the fairly thin (5–10 km) ocean-floor crust. Where does all the rock go? Emerging as hot dilated lava at the rift, the rock cools, contracts, and becomes denser, spreading out on both sides of the mid-ocean ridge. How can we know this? Natural radioactive isotopes determine the age of the rock at various places across the sea floor, and fossil magnetic compasses of magnetite and other crystals record the direction of the changing magnetic field of the Earth at the time when the rock congealed. These measurements show that the sea floor is expanding by several centimeters (a few inches) every year, fast enough to open up the Atlantic Ocean in a hundred million years. This sea-floor spreading is possible because at depths ranging from 70 to 250 km beneath the ocean-floor crust, temperatures are around 1,000 K and the layers are partially molten and plastic, so that they do not resist lateral motions of blocks of crust floating above them. These layers are called the asthenosphere, the force-free or weak zone.

The rocks of the continental plates are less dense than the basaltic sea-floor crust, and so they float higher above the asthenosphere. Drifting like “rockbergs” carried along by the expansion of the floor of the Atlantic, North America and Europe move steadily farther away from each other, as do South America and Africa. As noted earlier, there was no Atlantic Ocean in a not so distant past: all the continents were clustered in a single land mass, Pangaea. This supercontinent was torn apart some 200 million years ago by the outbreak of the Mid-Atlantic Ridge, giving birth to the Atlantic. Other separations followed: the Indian subcontinent split off from Antarctica 135 million years ago, Australia was cast adrift some 70 million years later. But what was the situation earlier than 200 million years ago? Is it likely that Pangaea had existed in splendid isolation ever since the cooling of the Earth’s crust billions of years before? There is some evidence that, on the contrary, other continents existed and drifted before then, meeting some 300 million years ago in a grand collision forming Pangaea. The sea floor is completely renewed every hundred million years or so, but some continental rocks are more than two billion years old. However, the continental blocks or plates can be broken, their pieces reassembled into new continents. As long as the Earth’s interior remains hot, mantle convection will recycle the ocean-floor crust. The rift, where the “new” ocean-floor crust appears, changes over the course of time, with volcanic activity dying out in one place and erupting somewhere else, breaking up a continent and giving birth to a new ocean.

The continents drift as if they were made of rigid plates; and this also appears to be the case of much of the spreading sea floor. But where then does all the continually created sea floor go? After all, the Earth is round! A moving plate must ultimately run into another plate. Basically, there are two possible outcomes. Either one of the plates passes under the other and in so doing it returns to the mantle in a subduction trench; or else the crust is compressed and folded in the collision, with formation and uplift of a mountain range. At the bottom of the Pacific Ocean, the Nazca plate, having emerged from the East-Pacific spreading ridge, slides under the American plate in the Peru-Chile subduction trench along the west coast of South America, and in so doing it pushed up the Andes. Further north, the little Juan de Fuca plate slides under the coast from California to Alaska. To the west, the Pacific (sea-floor) plate slides under the continental shelf and pushes up the Aleutian and Kuril Islands and Japan. Behind these islands are only shallow seas, above the continental shelf: the Bering and Okhotsk seas, the Sea of Japan, and the China Sea. In some places, one oceanic plate slides under another: thus the Pacific plate plunges under the Philippines plate in the (11,000 m or 36,000 ft. deep) Mariannes trench.

Although the motions are slow, subduction entails friction between the plates, and also stress, strain, and heat. The waterlogged sediments accompany the ocean-floor crust in the subduction, and the presence of water facilitates the melting of rocks at depths of several dozens or hundreds of kilometers. Sometimes the stress on the crust is relieved by sudden slippage or breakage, an earthquake; the release of heat and gas keeps volcanoes active. This happens all around the “ring of fire” of the Pacific, from the intermittently active volcanoes (Mount Lassen, Mount Saint-Helens) of California, Oregon, and Washington state, north and west to Alaska and the Aleutian Islands, then to Japan by way of the Kamchatka Peninsula and the Kuril Islands, south to Indonesia and further on to New Zealand where the Pacific plate sinks into the mantle. A plate can also slide along another plate, along a fault, sometimes getting stuck. If the sticking point gives way all at once, the sudden shift of the plate produces an earthquake, as in California along the San Andreas fault. Along the west coast of South America, the stick and slip of the Nazca plate as it passes under the American plate triggers frequent earthquakes.

The continental plate of North America finds itself compressed in a sort of vise between the spreading of the Atlantic on the east and that of the eastern Pacific on the west. Perhaps this explains the uplift of the western third of the continent, from the Pacific coast to the Rocky Mountains. Does it also explain the extreme violence of the earthquake that took place in 1811, near the center of the continent and far from subduction trenches, mid-ocean ridges, or any major fault? This quake (at New Madrid, on the Mississippi, about halfway between Memphis and St. Louis) would cause a major catastrophe today, but at the time, population density was small and most construction light. As the African plate approaches Europe, progressively closing up the Mediterranean, the Alps have been lifted up, along with sedimentary rocks and fossil seashells. The effect of compression appears much more spectacularly in the uplift of the Himalayas and the Tibetan plateau, result of the collision of the Indian subcontinent with the Eurasian plate that began 55 million years ago, continuing to this day. Another spectacular example: the Caucasus Mountains of Armenia, Azerbaijan, and Georgia, resulting from the collision of the Arabian plate with Eurasia. In these areas, the high risk of collapse of buildings in stone (or in Soviet concrete, as in Armenia in December 1988)¹³ makes quakes particularly murderous. Natural disasters are most disastrous when people are poorly prepared for them. It’s been said that “one learns geology the day after an earthquake,” but how many people remember the lesson?

Other dangerous encounters take place in the Caribbean, where the local Soufrière eruption buried a large part of the island of Montserrat in 1997, or around the Mediterranean, where a quake hit Assisi the same year, and where a city like Nice is at risk. Where volcanoes are involved, we are now able to avoid the enormous death tolls of past catastrophic eruptions, such as the eruption of Vesuvius (near Naples) that buried the cities of Pompeii and Herculaneum in 79 A.D., or the eruption of La Montagne Pelée at 8:02 A.M. on May 8, 1902, which buried the city of St. Pierre de la Martinique (28,000 dead, only 2 survivors). It is now possible to avoid such death tolls, thanks to progress in the understanding of volcanoes and in techniques for closely monitoring their swelling, the seismic activity specifically associated with them, and the changes in outflow from hot springs. Indeed, a wide range of phenomena announce an impending eruption, even if these signs cannot tell geologists everything. It was clear that Mount Saint-Helens (Washington state) was going to explode in May 1980, and the area was evacuated, but there still were victims, in part because it was not foreseen that most of the explosive force would emerge to the side rather than upward. Today, the number of casualties resulting from volcanic eruptions can be kept small if we have the will, provided that the necessary observing stations and alert systems are established and maintained in the regions at risk, provided also that scientists and technicians can convince the public and public safety officials of the reality and seriousness of the risk. This has not always been the case, for example when a terrible mudslide wiped out Armero (Colombia) following the eruption of the volcano Nevada del Ruiz on November 13, 1985. Will Naples have to be evacuated one day? Neither Mount Vesuvius nor the nearby Phlygrean Fields are extinct. Mount Popocatapetl (“smoking mountain” in the Nahuatl language), only 60 km (36 mi.) southeast of Mexico City, occasionally makes people nervous, and with reason. An enormous catastrophe was avoided when the Philippine authorities evacuated the vicinity of Mount Pinatubo before it erupted in June 1991.¹⁴ One of the factors contributing to this decision was the impressive documentary film put together by the Krafft couple. Volcano specialists from France, Katia and Maurice Krafft were among the 43 people who died when Mount Unzen (Japan) erupted on June 3, 1991. They knew the risks they were taking in staying close to the volcano.

On the seismic front, the search is on for ways to predict earthquakes. Along faults, regular gentle slippage of one plate relative to the adjoining one is probably the least dangerous course of events, and when monitoring shows that things are getting stuck, the red flags go up. Such monitoring can at least help risk assessment, even if it cannot reliably predict the day when a quake will take place. Some physicists, in particular the group in Greece known by their initials VAN, claim that certain electromagnetic wave signals can be detected just before a quake strikes. Whatever the merit of this particular method, it makes sense to look for advance signals of the sudden slippage or rupture under stress that constitutes the physical process of an earthquake. Nevertheless, the methods of prediction so far proposed are highly controversial: none is generally applicable, and even those for specific areas remain at best highly uncertain. Several million people were evacuated from the city of Haicheng (China) hours before a predicted earthquake struck in February 1975, but there was no usable prediction of the August 1976 T’ang-shan earthquake that took 700,000 lives. In any event, any false alarm, for example in Los Angeles, could have catastrophic consequences. And while people may be able to flee from threatened cities, their homes and workplaces cannot. Antiseismic construction codes and methods have proven their worth to some extent, but there still have been bad surprises, for example in Oakland, California (1989) and in Kobe, Japan (1995). It is hard to imagine any public or corporate decision to limit population growth and economic development in such high-risk regions as California, central Japan, or Istanbul, and so the highest priority must be given to making buildings more earthquake resistant and to teaching people how to react when a quake does strike.

Effective earthquake prediction remains elusive, but the international scientific community has succeeded in setting up an effective tsunami alert system, at least around the Pacific Ocean. The word tsunami comes from the Japanese, and it stands for big waves that wash over coastal areas. They are caused by underwater earthquakes or mudslides, and not by tides, although they are sometimes called “tidal waves.” The wave is triggered by the sudden shock, shift, or vibration of the ocean floor, and it then propagates in all directions at a speed in the range 700–1,000 kilometers per hour (400–700 mph). In the open ocean, the wave, only a few centimeters (1–2 in.) high with wave crests about 100 km (60 mi.) apart, cannot be seen. The wave continues its travel for thousands of kilometers, but with decreasing ocean depth close to the coast, wave height can grow to tens of meters. Triggered by an earthquake off the coast of Chile on May 22, 1960, causing thousands of casualties in Chile and Peru, a tsunami reached Hawaii fifteen hours later, wiping several coastal villages off the map and partly destroying the city of Hilo; six hours later still, the same tsunami reached Japan, causing hundreds of casualties. On August 27, 1883, the volcano Krakatoa (Indonesia) exploded with a force of 100 megatons of TNT (as big as or bigger than any H-bomb ever tested): this triggered a local tsunami with waves 40 meters (130 ft.) high, drowning 36,000 persons.¹⁵ For local tsunamis, there is not enough time for warning, but an alert can be given wherever the tsunami has to travel far from the initial shock. Seismological measurement stations the world over immediately inform the International Tsunami Warning Center outside Honolulu of the location and strength of any seismic tremor likely to trigger a tsunami. The center computes the tsunami’s travel times to the different Pacific coasts and issues the warning. Given the size of the Pacific, those travel times can be several hours long, and effective precautions can be taken. Elsewhere—for example, in the Caribbean or the Mediterranean—advance warning time is much shorter. In 1908 a Mediterranean tsunami drowned thousands in Messina (Sicily) and Reggio di Calabria. If you are near the seashore and have the impression that the sea is suddenly emptying out and retreating far far away, you had best head for the hills, or at least higher country, as fast as you can, because the sea can come back in an enormous wave that can wash everything away over hundreds of yards.

For the most part, earthquakes and volcanic eruptions occur either along the mid-ocean ridge, or along the lines where plates collide or slip past one another. These tremors and exhalations are the superficial signs of the convective currents that stir up the mantle, with magma rising at the mid-ocean ridge and ocean-floor crust descending in the subduction trenches. Other volcanoes and other earthquake sites are to be found at what are called “hot spots,” isolated points where magma comes up through the crust. One of these is the Kilauea caldera near the southeast coast of the big island of Hawaii. Tourists can leave an orchid there, at the edge of the vast Halema’uma’u crater, in homage to the goddess Pele. And if the continuing exhalations of sulfur dioxide (SO₂) and of hydrochloric acid (HCl) do not bother them too much, they can descend into the crater, taking care not to stray from the marked trail because in other places, the recently solidified glassy crust can break under their weight. The crater was full of molten lava when Mark Twain visited it in June 1866, and it could fill again in a coming year. Lava is emerging today from the Pu’u O’o crater, a bit closer to the sea. The lava flow cut the road and buried the Waha’ula temple built some eight centuries ago, and it plunges into the sea, generating an enormous cloud of steam laden with acid vapors and glassy dust. Off the coast, 32 km (20 mi.) to the southeast, a new island already given the name Loihi is being born: its peak is still a thousand meters below sea level, but it will emerge in 50,000 or 100,000 years. Returning to the center of the big island, one can climb the slopes of Mauna Loa, a volcano that since 1960 has been relatively quiescent. But Mauna Loa is not extinct, and in 1942 a lava flow threatened the city of Hilo. Further to the north, astronomers often visit Mauna Kea (“white mountain” in Hawaiian). With an altitude of 4,200 meters (14,000 ft.), the peak is often snow-covered. Well-equipped observatories have been established there to take advantage of the clean thin air, funded by many different countries (in particular the Canada-France-Hawaii 360-cm or 142-inch telescope). Obviously, the astronomers hope that the volcano is extinct. Both Mauna Loa and Mauna Kea rise more than 8,000 meters above the nearby ocean bottom. Mau’i with Haleakala, Oahu with Diamond Head, indeed all the islands of the Hawaiian chain out to Midway 2,000 km (1,200 mi.) west of Honolulu, mark both the volcanic activity of the past and the spreading of the Pacific ocean floor. What seems to have happened is that a fixed spot has from time to time heated up and erupted, forming one or more volcanoes, building up a new island. And then the motion of the sea floor has slowly but inexorably carried the island off to the west, removing the volcano from the “hot spot” and thus extinguishing it. With the next spurt of hot-spot activity, a new volcano is born, a new island built up. Elsewhere, in the Pacific, in the Indian Ocean, and in the Atlantic from São Tomé to Mount Cameroon, strings of volcanic islands mark the spurts of activity of hot spots under a moving ocean floor. Other hot spots exist under the continents—for example, the one that heats up the geysers and hot springs of Yellowstone.

From the mouths of a thousand volcanoes, the Earth spews out lava, steam, carbon dioxide, sulfur dioxide, hydrogen sulfide (H₂S), hydrochloric acid, and more. In a distant past, fits of volcanic frenzy buried vast areas with lava, in the Deccan Traps of India, and the Columbia River basalt of the northwest United States. On the continents, such solidified lava discharges last long, but on the ocean floor they are recycled in the mantle by way of the subduction trenches. The lava emerging today in the rift or in newly opened craters on Hawaii or Montserrat may contain minerals already present in the sea floor of ancient oceans. The steam spouted by volcanoes today comes for the most part from recycled water that was included in the sediments descending in the subduction trenches, water that was for a time lost to the oceans and atmosphere. And so, even though water vapor has emerged from volcanoes for billions of years, the seas are not full. From the ocean bottom through the sea floor, water is drawn into a still deeper water cycle, reappearing in more or less vigorous volcanic emanations. But apart from geysers, this subsurface cycle can only supply the sources of rivers by way of the atmospheric branch of the water cycle. Some of the steam emerging from Pu’u O’o on Hawaii may conceivably fall as rain on the Kahlua coffee plantations of the western slope of the island the same evening, rejoining the waters of the Pacific by week’s end. Each raindrop of the showers that water the apple orchards of Normandy includes a few H₂O molecules coming from the Soufrière, Etna, and Kilauea volcanoes, among billions of billions of water molecules evaporated from the surface of the Atlantic Ocean. And those billions of billions may include a few molecules recently emerged from the mid-Atlantic ridge. Some of that water will be incorporated in the apple that you eat on a visit to Paris. The rest will drain into the English Channel, the Atlantic Ocean, or the North Sea, unless it is reevaporated. What does evaporate can condense to form a cloud, and a few days later may water the Tokay (Tokaj) vineyards in Hungary and run off in the Danube River to the Black Sea. Whenever you eat an apple or drink a glass of wine, you are absorbing water that has cycled through the atmosphere thousands of times since you were born. But you are also absorbing some water molecules that have only been out in the open air for a few days or weeks, after tens or hundreds of millions of years beneath the Earth’s crust.