12. The Last Hurrah! of Classical Science, The Scientists

The last great triumph of classical science did depend on one discovery which, with hindsight, belongs in the twentieth century post-classical world (‘post-classical’ in the scientific, not the literary or art-historical, sense, meaning that it is based on relativity theory and quantum mechanics). That was the discovery of radioactivity (itself made in the nineteenth century), which provided a source of heat that could prevent the interior of the Earth from cooling into a solid, inert lump on the sort of timescales required by the uniformitarian ideas developed by Lyell and his predecessors. It would take the theories of relativity and quantum physics to progress from the discovery of radioactivity to an explanation of the phenomenon and an understanding of how the conversion of mass into energy keeps the stars shining. But, just as Galileo could study the way pendulums swing and balls roll down inclined planes without knowing how gravity worked, all that geophysicists needed to know about radioactivity was that it did provide a way to keep the Earth warm inside – that there was a source of energy to drive the physical processes which have shaped the surface of the planet over an immense span of time and continue to do so today. Armed with that knowledge they could develop geology into geophysics, explaining the origin of the continents and ocean basins, the occurrence of earthquakes, volcanoes and mountain building, the wearing away of land by erosion and much more besides; all in terms of the kind of science that would have been well understood by Isaac Newton or Galileo Galilei, let alone William Thomson or James Clerk Maxwell.

In spite of the importance of Lyell’s influence (especially in the English-speaking world, and most especially on Charles Darwin), you shouldn’t run away with the idea either that uniformitarianism swept the board after the publication of his Principles of Geology or that most geologists of the nineteenth century actually cared all that much about the debate concerning the physical causes which had shaped the globe. Indeed, you couldn’t really say that there was a debate; different people put forward different models, each of which had their adherents, but the rivals didn’t meet to discuss the merits of their rival models or engage much in any kind of confrontation in print. The first task, still very much at the forefront throughout the nineteenth century, was to carry out the field work which put the strata in order and gave geologists a relative timescale to work with, so that they knew which rocks were older and which were younger. As far as investigating ideas about the origin of those strata went, there were even shades of uniformitarianism, and it was widely thought that although the same kinds of forces had been at work in the past as today (earthquakes and volcanoes, for example), they may have been more powerful in the past, when the Earth was younger and presumed to have been hotter. Lyell’s uniformitarianism said that continents could be converted into sea floor and ocean floor raised up to make continents; but another (still uniformitarian) school of thought, known as permanentism, held that continents had always been continents and oceans had always been oceans. The permanentists were particularly strong in North America, where James Dana (1850–1892), professor of natural history and geology at Yale University, was its leading advocate. He linked the hypothesis with the (not unreasonable, given the state of knowledge at the time) idea that the Earth was gradually shrinking, contracting as it cooled, and that mountain ranges such as the Appalachians were produced, in effect, by the wrinkling of the Earth’s crust as it shrank.

In Europe, the idea of contractionism was developed along different lines, as a variation on catastrophism. This idea culminated in the final decades of the nineteenth century in a synthesis of older ideas developed by Eduard Suess (1831–1914), who was born in London (the son of a German wool merchant) but moved with his family as a child first to Prague and then to Vienna, where he eventually became professor of geology at the university. Suess’s model saw contraction as the driving force for rapid bursts of dramatic change, separated by long intervals of relative calm, on a cooling and contracting Earth. He suggested that the present-day land masses of Australia, India and Africa were fragments of a much greater land mass (which he dubbed Gondwanaland, after a region of India) that had once existed in the southern hemisphere, much of which had sunk into the cooling interior. The wrinkling crust of the Earth had, in this picture, formed folds (mountain ranges and rifts) and large chunks (such as the Atlantic, as well as in the southern hemisphere) had subsided into space made available in the interior as it cooled and contracted, forming new ocean basins between formerly connected land masses; but this happened in sudden bursts, not as a gradual, continuing process. The model failed to stand up to proper investigation. For example, the amount of wrinkling and folding required to produce the Alps alone, squeezing (according to the Suess synthesis) 1200 kilometres of crust into 150 kilometres of mountains, corresponded to a cooling of 1200 °C. Even greater cooling would be required to produce the shrinking alleged to have given rise to the Himalayas, the Rockies and the Andes, which formed at essentially the same time as the mountains of the Alps. But the key blow to all such models was the discovery of radioactivity, made at almost the same time Suess was developing his synthesis, which showed that the Earth’s interior was not, in fact, cooling dramatically at all. The story of Suess’s synthesis is, though, significant for two reasons. First, it highlights the lack of any ‘standard model’ of Earth history at the beginning of the twentieth century; second, it gave us a name, Gondwana, which would become familiar as the idea of continental drift became established. But although that idea itself had also been aired in the nineteenth century, it would not become established until well into the second half of the twentieth century – less than fifty years ago.

Among the variations on the theme of continental drift put forward in the nineteenth century there was the idea that the continents might be sitting on magnetized crystalline foundations and were being swept northwards by a magnetic flow, and the suggestion that the Earth had originally been not only smaller than it is today but tetrahedral in shape, with the continents originally nestling close together but being ripped apart from one another in a catastrophic expansion event which also flung the Moon out of the Mediterranean basin and into its orbit. In 1858 (the year before the publication of the Origin of Species), Antonio Snider-Pellegrini, an American working in Paris, published a book, La Création et ces mystères devoilés, which put forward a bizarre model based on his interpretation of the Bible. This involved a series of catastrophes taking place on a rapidly shrinking Earth at the beginning of its history. It is only worth mentioning because the book marked the first publication of a map bringing together the continents on both sides of the Atlantic Ocean, which was used to explain the similarities between fossils found in coal deposits on opposite sides of the ocean. The map has been widely reprinted, giving the misleading impression that Snider-Pellegrini actually had a sensible model of continental drift. A somewhat more scientific (but still catastrophic) version of continents in motion was raised by Osmond Fisher in a paper published in the science journal Nature on 12 January 1882. He took up an idea proposed by the astronomer George Darwin (1845–1912; one of Charles Darwin’s sons) that the Moon had formed when the young Earth split into two unequal parts. Fisher suggested that the Pacific basin marked the wound where the Moon had been torn out of the Earth, and that continental material on the other side of the world would have cracked and the fragments been pulled apart as the remaining surface of the Earth slowly moved in the direction of the hole as it began to fill in.

In the first decades of the twentieth century, other versions of continental drift were also proposed. But the one which (eventually) made its mark and influenced the development of the Earth sciences was the one put forward by the German meteorologist Alfred Wegener, initially in 1912. Coming from a different scientific discipline (he originally trained as an astronomer), Wegener seems to have known little of the plethora of older ideas about continental drift (probably just as well, seeing how harebrained some of them were). His ideas became influential not just because he developed a more complete model than those predecessors, but because he campaigned for it over a period of decades, seeking out more evidence in support of his idea, defending the model in the light of criticism and publishing a book which ran into four editions before his untimely death in 1930. Wegener kicked up a fuss about continental drift, rather than just publishing his ideas and leaving them to sink or swim on their own. Although many of his detailed ideas were incorrect, his overall concept has stood the test of time, and Wegener is now rightly regarded as the father of the theory (as it now is) of continental drift.

Wegener was born in Berlin on 1 November 1880, and studied at the universities of Heidelberg, Innsbruck and Berlin, obtaining his doctorate in astronomy from Berlin in 1905. He then joined the Prussian Aeronautical Observatory at Tegel, where he worked for a time alongside his brother Kurt (literally alongside on one occasion, when the brothers undertook a balloon flight lasting 52½ hours, a record at the time, to test instruments). From 1906 to 1908 Wegener worked as meteorologist to a Danish expedition to the interior of Greenland, and on his return he joined the University of Marburg as a lecturer in meteorology and astronomy. He published a meteorological textbook in 1911, but by then was already developing his ideas on continental drift, which first appeared in print in 1912 in a pair of papers based on talks he had given in Frankfurt am Main and Marburg in January that year. As Wegener later recalled, in 1910 one of his colleagues at Marburg had been given a new world atlas, and while looking at it Wegener was struck (like others before him) by the way the east coast of South America and the west coast of Africa looked as if they ought to fit together, like pieces in a jigsaw puzzle, as if they had once been joined. Although intrigued, he regarded the idea as improbable and didn’t take it forward until the spring of 1911, when he came across a report discussing the paleontological similarities between the strata of Brazil and Africa. The evidence was presented in that report in support of the idea of a former land bridge linking the two continents; but Wegener saw things differently. As he wrote in the first edition of what became his masterwork, Die Entstehung der Kontinente und Ozeane, published in 1915:¹

This induced me to undertake a cursory examination of relevant research in the fields of geology and palaeontology, and this provided immediately such a weighty corroboration that a conviction of the fundamental soundness of the idea [of continental drift] took root in my mind.

One other piece of evidence helped to persuade Wegener that he was on to something – the jigsaw-puzzle-like fit of the continents is even better if they are matched up not along the present-day shorelines (which depend on the height of the ocean today), but along the edges of the continental shelf, the true edge of the continents, where there is a steep plunge down to the ocean floor. But although the idea had taken root, several distractions delayed its full flowering. Shortly after the presentation of his first drift ideas in those talks in January 1912, Wegener set out on another Greenland expedition, returning in 1913 and marrying Else Köppen.² Any plans they had for a quiet academic life were shattered by the First World War, in which Wegener was called up as a reserve lieutenant and served on the Western Front, where he was wounded twice in the early months; unfit for further active service, he worked in the meteorological service of the army after his recovery. It was while on convalescent leave that he wrote the first version of his famous book (whose title translates as The Origin of Continents and Oceans). This made very little impact at the time. It was published in 1915, at the height of the war, and was scarcely more than a pamphlet, only running to 94 pages. After the war, Wegener worked for the German Marine Laboratory in Hamburg (again alongside his brother), and was also a lecturer in meteorology at the then-new University of Hamburg. He established a reputation as a distinguished meteorologist, but also continued to work on his model of continental drift, producing new editions of his book (each larger than its predecessor) in 1920 and 1922. Friends worried that this might damage his reputation, but whatever people thought about the drift idea, Wegener was such a good meteorologist that in 1924 he was appointed professor of meteorology at the University of Graz, in Austria. The same year, he published (with Wladimir Köppen) the first attempt at an explanation of past climates based on continental drift, and both French and English translations of the third (1922) edition of Die Entstehung appeared. But just when Wegener seemed to be gaining an audience for his ideas, the opportunity to promote them further was snatched from him, although he did prepare a fourth edition of his book, responding to criticisms of the third edition from the English-speaking world that had now been introduced to his ideas, which was published in 1929. In 1930, Wegener set off on yet another Greenland expedition, this time (at the age of 49) as its head; the aim of the expedition was to gather evidence in support of the drift hypothesis. The expedition ran into trouble on the desolate Greenland ice cap, and with supplies running low at an inland camp, on 1 November 1930 (his fiftieth birthday) Wegener set off for the main base on the coast with an Inuit companion. He never made it. The following spring, his body was found on the ice cap on the route between the two camps, neatly wrapped in his sleeping bag and marked with his upright skies; his companion was never seen again. Now, the continental drift idea would have to sink or swim without the aid of its chief proponent.

Wegener’s model envisaged the Earth as made up of a series of layers, increasing in density from the crust to the core. He saw that the continents and the ocean floors are fundamentally different, with the continents explained as blocks of light granitic rock (known as sial, from the name silica–alumina which describes their composition) essentially floating on denser basaltic rock (sima, from silica–magnesium), which (underneath a layer of sediment) forms the rock of the ocean floor. He said that the present-day continental blocks still have essentially the same outlines as they have had since the breakup of a single supercontinent, Pangea, which contained all the land surface of the planet at the end of the Mesozoic era (about 150 million years ago by modern dating). One big weakness of Wegener’s model is that he had no reason for the breakup of Pangea and could only invoke rather vague ideas such as a ‘retreat from the pole’ caused by centrifugal forces, or possible tidal effects, to produce continental drift. But he went further than his predecessors in pointing to the sites of rift valleys (such as the East African rift valley) as locations of incipient continental breakup, indicating that whatever the process is that drives continental drift, it is still continuing today, and thereby making his version of the drift idea a uniformitarian one. Crucially, he also based his ideas on an Earth of constant size, with no catastrophic (or even gradual) contraction or expansion. One of the weakest features of the model is that Wegener envisaged the continents ploughing through the sima of the sea floor, which geologists (rightly) found hard to swallow. But he linked his ideas with the way mountains had formed along the eastern edges of the North and South American continents as they had drifted away from Europe and Africa, with the continents being crumpled up as they ploughed through the sima. Mountain ranges such as the Himalayas, in the hearts of land masses, could be explained by the collision of continents.

Like the curate’s egg, the details of Wegener’s hypothesis were good in parts. Where it was particularly good was in the evidence he marshalled from paleoclimatology, showing how glaciation had occurred in the distant past simultaneously on continents that are now far apart from one another, and far from the polar regions; where it was particularly bad (apart from the fact that he often ignored evidence which did not support his case, which made geologists suspicious of the whole package) was in his belief that continental drift was happening so rapidly that Greenland had broken away from Scandinavia only 50,000 to 100,000 years ago, and was moving westward at a rate of 11 metres per year. This suggestion came from geodetic surveys carried out in 1823 and 1907, and the measurements were simply inaccurate; today, using laser range finding with satellites, we know that the Atlantic is actually widening at a rate of a couple of centimetres per year (it was, incidentally, in pursuit of improved geodetic data that Wegener made his last, fatal, trip to the Greenland ice cap). But the most valuable contribution he made to the development of the idea of continental drift was his synthesis, gathering evidence to support the former existence of the supercontinent of Pangea by linking mountain ranges, sedimentary rocks, evidence from the scars of ancient glaciations, and the distribution of both fossil and living plants and animals. In a telling analogy, Wegener made a comparison with a sheet of printed paper torn into fragments. If the fragments could be reassembled so that the printed words joined up to make coherent sentences, it would be compelling evidence that the fit of the pieces was correct; in the same way, the kind of evidence he gathered formed a coherent geological ‘text’ when the pieces of Pangea were reassembled. It is this broad sweep of evidence which made the case for continental drift, even before the mechanisms were fully understood.

In fact, even the key component of the mechanism of continental drift was already in place by the end of the 1920s, as one geologist in particular appreciated. That geologist was Arthur Holmes (1890–1965), who by the 1920s had become a leading expert on radioactive decay and was at the forefront of efforts to measure the age of the Earth using radioactive techniques. More than any other individual, he was, indeed, ‘the man who measured the age of the Earth’. Holmes came from an unremarkable family in Gateshead, in the northeast of England (his father was a cabinet maker and his mother worked as a shop assistant). He went to the Royal College of Science in London in 1907, after passing examinations for a National Scholarship which provided him with thirty shillings a week (£1.50) during the academic year. This was not enough to live on even in 1907, and there was no prospect of financial support from his parents; Holmes just had to make do as best he could.

Around this time, both radioactivity and the age of the Earth were hot scientific topics, and the American Bertram Boltwood (1870–1927) had recently developed the technique for dating samples of rock from the proportions of lead and uranium isotopes they contain. Since the radioactive decay of uranium eventually produces lead, with (as we shall see in Chapter 13) a characteristic timescale, measuring these ratios can reveal the ages of rocks. As his undergraduate project in his final year, Holmes used the technique to date samples of Devonian rock from Norway, coming up with an age of 370 million years. Scarcely ten years into the twentieth century, even an undergraduate could get a date for a piece of rock, which was clearly by no means the oldest rock in the Earth’s crust, that was far in excess of the timescale for the Solar System allowed by the idea of heat being released from the Sun only as a result of its gravitational collapse. Graduating in 1910 with a glowing reputation but a burden of debt from his undergraduate days, Holmes was delighted to obtain a well-paid post for six months as a prospecting geologist in Mozambique, at £35 per month. A serious bout of blackwater fever delayed his return home, and he also contracted malaria (a blessing in disguise, since it prevented him from joining the army in the First World War). With his finances now in order (he made a profit on the trip of £89 7s 3d), Holmes was able to join the staff of Imperial College (as the Royal College of Science had metamorphosed into in 1910), where he stayed until 1920, receiving a doctorate in 1917. He then worked in Burma for an oil company, returning to Britain in 1924 to become professor of geology at Durham University. He moved on to the University of Edinburgh in 1943, and retired in 1956. By then, he had firmly established the radioactive technique for measuring the ages of rocks, coming up with an age for the Earth itself of 4,500 ± 100 million years.³ Along the way, he produced an influential textbook, Principles of Physical Geology (the title deliberately chosen as a nod to Lyell), which was first published in 1944 and, in revised editions, has been a standard text ever since. Part of its success may be explained by the way Holmes tackled the task of making geology intelligible. As he later wrote to a friend, ‘to be widely read in English-speaking countries think of the most stupid student you have ever had then think how you would explain the subject to him’.⁴

Holmes’s interest in continental drift was almost certainly aroused before 1920 by one of his colleagues at Imperial, John Evans, who read German fluently and became an early enthusiast for Wegener’s ideas (he later wrote the Foreword for the first English edition of Wegener’s book). The third edition of the book had just been published in England when Holmes returned from Burma, and this seems to have been the stimulus for him to take up the idea, during a break from the uranium–lead work, as soon as he had established himself in Durham. Although he started out favouring the contraction hypothesis, his understanding of radioactivity and the potential this provided to generate heat inside the Earth soon led him to change his views. The idea that convection might be associated with mountain building and continental drift was planted in his mind by the discussion of them in the Presidential Address given by A. J. Bull to the Geological Society in London in 1927 (just a hundred years after Charles Darwin went up to Cambridge University intending to become a clergyman). In December that year Holmes presented a paper to the Edinburgh Geological Society which took up these ideas. He suggested that although the continents indeed floated on denser material, more or less as Wegener proposed, they did not move through the sima. Rather, this denser material itself moved around very slowly, stirred by convection currents produced by the heat within the Earth, cracking apart in some places (such as the ocean ridge down the centre of the Atlantic Ocean) and pushing the continents on either side of the crack apart, while they collided in other parts of the globe. Apart from the radioactive heating, the key component of Holmes’s model was time – ‘solid’ rock, warmed from beneath, could indeed stretch and flow, like very thick treacle (or like the ‘magic putty’ you can find in some toy shops), but only very slowly. It is no surprise that one of the first geologists to espouse continental drift was one of the first people to appreciate quantitatively the immense age of the Earth, and to be actively involved in measuring it. In 1930, Holmes produced his most detailed account of continental drift, describing how convection currents operating inside the Earth as a result of heat generated by radioactive decay could have caused the breakup of Pangea, first into two large land masses (Gondwanaland in the southern hemisphere; Laurasia in the north), which in turn fragmented and drifted to form the pattern of land that we see on the surface of the Earth today. All this was published in the Transactions of the Geological Society of Glasgow, including an estimate very much in line with present-day measurements that the convection currents would move continents about at a rate of some 5 centimetres a year – enough to produce the Atlantic basin from a crack in the crust over an interval of about 100 million years.

Very many pieces of the modern version of continental drift were already in place in 1930, and Holmes presented the evidence for drift in the final chapter of his Principles in 1944, clearly arguing the case, but honestly pointing out the flaws in Wegener’s own presentation:

Wegener marshalled an imposing collection of facts and opinions. Some of his evidence was undeniably cogent, but so much of his advocacy was based on speculation and special pleading that it raised a storm of adverse criticism. Most geologists, moreover, were reluctant to admit the possibility of continental drift, because no recognized natural process seemed to have the remotest chance of bringing it about…Nevertheless, the really important point is not so much to disprove Wegener’s particular views as to decide from the relevant evidence whether or not continental drift is a genuine variety of earth movement. Explanations may safely be left until we know with greater confidence what it is that needs to be explained.

And right at the end of his chapter on continental drift, after making the case for convection as the driving force for the process, Holmes wrote:

It must be clearly recognized, however, that purely speculative ideas of this kind, specially invented to match the requirements, can have no scientific value until they acquire support from independent evidence.

I wonder if Holmes knew how closely he was echoing the words put into the mouth of his fictional namesake by Arthur Conan Doyle in A Scandal in Bohemia:

It is a capital mistake to theorise before one has data. Insensibly, one begins to twist facts to suit theories, instead of theories to suit facts.

Virtually nothing had happened to strengthen the case for continental drift between 1930 and 1944 precisely because there were no new facts to go on. Of course, there was some resistance among the old guard to the new ideas, just because they were new – there are always people reluctant to throw out everything they have been taught in order to espouse a new understanding of the world, no matter how compelling the evidence. But in the context of the 1930s and 1940s, the evidence in support of continental drift was persuasive (maybe very persuasive, if you took on board Holmes’s work) rather than compelling. There were still other well-regarded rival ideas, notably permanentism, and with Wegener dead and Holmes concentrating on his dating techniques, nobody went in to bat for continental drift, which gradually fell out of what favour it had had (to the point where just about the only criticism of Holmes’s great book came from people who said that he should not have included a chapter supporting such cranky ideas). What made continental drift first respectable, and then an established paradigm, the standard model of how the Earth works, was indeed new evidence – new evidence that emerged in the 1950s and 1960s thanks to new technology, itself developed partly as a result of the dramatic boost to all the technological sciences provided by the Second World War. This is also the first example we shall encounter in the present book of the way science became a discipline where real progress could only be achieved by large numbers of almost interchangeable people working on big projects. Even a Newton could not have obtained all the information needed to make the breakthrough that converted the continental drift hypothesis into the theory of plate tectonics, although he undoubtedly would have been able to put the evidence together to form a coherent model.

Although the technological advances stemming from the Second World War eventually helped to provide the key evidence in support of continental drift, during the 1940s many geologists were working on war-related projects, serving in the armed forces or living in occupied countries where there was little opportunity for global scientific research. In the immediate aftermath of the war, rebuilding in Europe and the dramatically changed relationship between science and government in the United States helped to delay the development and application of the new techniques. Meanwhile, although papers discussing drift (both for and against) were published, it largely remained a backwater of the geological sciences. The idea was, however, ready and waiting in the wings when that new evidence, that might otherwise have proved extremely puzzling and difficult to explain, began to come in.

The first new evidence came from the study of fossil magnetism – the magnetism found in samples of rock from old strata. The impetus for this work itself came originally from the investigation of the Earth’s magnetic field, whose origin was still a puzzle in the 1940s. Walter Elsasser (1904–1991), one of many German-born scientists who left Germany when Adolf Hitler came to power and ended up in the United States, began, in the late 1930s, to develop the idea that the Earth’s magnetism is generated by a natural internal dynamo, and he published his detailed ideas almost as soon as the war ended, in 1946. The idea was taken up by the British geophysicist Edward Bullard (1907–1980), who had worked during the war on techniques for demagnetizing ships (‘degaussing’) to protect them from magnetic mines. In the late 1940s, Bullard was working at the University of Toronto, where he developed further the model of the Earth’s magnetic field as the product of circulating conducting fluids in the hot fluid core of the planet (crudely speaking, convection and rotation in molten iron). In the first half of the 1950s, as Director of the UK National Physical Laboratory in London, he used their early electronic computer for the first numerical simulations of this dynamo process.

By that time, measurements of fossil magnetism had shown that the Earth’s magnetic field had had the same orientation relative to the rocks for the past 100,000 years. The rocks are magnetized when they are laid down, as molten material flowing from volcanoes or cracks in the Earth’s crust, and once set they preserve the pattern of the magnetic field in which they formed, becoming like bar magnets. But British researchers in particular (notably small groups based at the universities of London, Cambridge and Newcastle upon Tyne) had found that in older rocks the direction of the fossil magnetism could be quite different from the orientation of the present-day geomagnetic field, as if either the field or the rocks had shifted position after the strata solidified. Even more strange, they found that there seemed to be occasions in the geological past when the geomagnetic field had the opposite sense to that of today, with north and south magnetic poles swapped. It was this paleomagnetic evidence that made the debate about continental drift hot up at the beginning of the 1960s, with some people using the magnetic orientations of the rocks from particular times in the geological past as the ‘lines of print’ to be matched up across the joins of continental reconstructions, and finding that those reconstructions broadly matched the ones made by Wegener.

Alongside all this, there had been a huge development in knowledge of the sea bed, which makes up two-thirds of the crust of the Earth’s surface. Before the First World War, this was still largely a mysterious and unexplored world. The need for ways to counter the submarine menace encouraged development of the technology for identifying what lay beneath the surface of the ocean (particularly echo-location, or sonar) and the incentive to use the technology not just for detecting submarines directly but, after the war, to map the sea bed, partly out of scientific curiosity but also (as far as governments holding the purse strings were concerned) to locate hiding places for submarines. It was this technology which had begun, by the end of the 1930s, to fill in the outline features of sea floors, most notably indicating the presence of a raised system, a mid-ocean ridge, not just running down the Atlantic Ocean, but also forming a kind of spine down the centre of the Red Sea. The Second World War saw a huge improvement in the technology used for this kind of work, and the Cold War encouraged a continued high level of funding, as nuclear armed submarines became the primary weapons systems. In the United States, for example, the Scripps Institution of Oceanography had a budget of just under $100,000 in 1941, employed 26 staff and owned one small ship. In 1948, it had a budget of just under $1,000,000, a staff of 250 and four ships.⁵ What the resources of Scripps and other ocean researchers found was quite unexpected. Before the 1940s, geologists had assumed that the sea floor represented the most ancient crust of the Earth – even supporters of continental drift thought this way. Because they were assumed to be ancient, the sea floors were also assumed to be covered with huge amounts of ancient sediment worn off the land over the eons, forming an essentially featureless layer perhaps 5 or 10 kilometres thick. And the crust itself, beneath the sediment, was assumed to be tens of kilometres thick, like the crust of the continents. When samples were obtained from the ocean bed and surveys were carried out, they showed that all of these ideas were wrong. There is only a thin layer of sediment, and hardly any at all away from the edges of the continents. All of the rocks of the sea floor are young, with the youngest rocks found next to the ocean ridges, which are geologically active features where underwater volcanic activity marks the line of a crack in the Earth’s crust (so some of the rocks there have literally been born yesterday, in the sense that that is when they solidified from molten magma). And seismic surveys showed that the thickness of the Earth’s crust is only about 5–7 kilometres under the oceans, compare with an average of 34 kilometres for the continental crust (in places, the continental crust is 80 or 90 kilometres thick).

The pieces of the puzzle were put together in a coherent fashion by the American geologist Harry Hess (1906–1969), of Princeton University, in 1960. According to this model, which goes by the name of ‘sea-floor spreading’,⁶ the ocean ridges are produced by convection currents in the fluid material of the mantle (the layer of treacly rock just below the solid crust) welling up from deeper below the surface. This warm material is not liquid in the sense that the water of the oceans themselves is liquid, but is hot enough to flow slowly as a result of convection, a little like warm glass.⁷ The volcanic activity associated with the ocean ridges marks the place where this hot material breaks through to the surface. It then spreads out on either side of the ridge, pushing the continents on either side of the ocean basin apart, with the youngest rocks solidifying next to the ridges today and with the older rocks, laid down tens or hundreds of millions of years earlier, further away from the ridges, where they have been pushed to make room for the new material. And there is no need for the continents to be ploughing through the ocean crust – which is just as well, since surveys of the sea floor show no evidence of this. New oceanic crust created in this way is widening the Atlantic at a rate of about 2 centimetres a year, roughly half the speed that Holmes suggested. There are some echoes of Holmes’s ideas in Hess’s model, but the crucial difference is that where Holmes could only talk in general terms based on the fundamental laws of physics, Hess had direct evidence of what was going on and could put numbers derived from measurements of the ocean crust into his calculations. Holmes largely ignored the ocean basins in his model, for the very good reason that very little was known about them at the time; after Hess’s work had been fully assimilated, which took most of the 1960s, the ocean basins were seen as the sites of action in continental drift, with the continents themselves being literally carried along for the ride as a result of activity associated with the crust of the ocean floors.

Although the Atlantic is getting wider, this does not mean that the Earth is expanding at the rate required to explain the formation of the entire Atlantic basin in a couple of hundred million years, roughly 5 per cent of the age of the Earth, the rate required by those measurements. Convection currents go up in some places, but down in others. The second key component of Hess’s model of sea-floor spreading was the suggestion that in some parts of the world (notably along the western edge of the Pacific Ocean), thin oceanic crust is being forced down under the edges of thicker continental crust, diving back down into the mantle below. This explains both the presence of very deep ocean trenches in those parts of the world and the occurrence of earthquakes and volcanoes in places such as Japan – islands like those of Japan, indeed, are explained as being entirely produced by the tectonic activity⁸ associated with this aspect of sea-floor spreading. The Atlantic Ocean is getting wider, but the Pacific Ocean is narrowing. Eventually, if the process continues, America and Asia will collide to form a new supercontinent; meanwhile, the Red Sea, complete with its own spreading ridge, marks the site of a new region of upwelling activity, cracking the Earth’s crust and beginning to splinter Africa away from Arabia to the east.

As the model developed, it was also able to explain features such as the San Andreas Fault in California, where the widening of the Atlantic has pushed America westwards to overrun a less active spreading zone that formerly existed in what was, hundreds of millions of years ago, an even wider Pacific basin. Faults like the San Andreas also provided circumstantial evidence in support of the new ideas, as some geologists were quick to point out. There, blocks of the Earth’s crust are moving past one another at a rate of a few centimetres per year, roughly the same speeds required for the new version of continental drift and proof that the ‘solid’ Earth was by no means fixed in one permanent geographical pattern. The traditional analogy, which has never been bettered, is that sea-floor spreading is like a slow conveyor belt, endlessly looping round and round. Over the entire surface of the globe, everything cancels out and the planet stays the same size.⁹

Hess’s model, and the evidence on which it was based, inspired a new generation of geophysicists to take up the challenge of building a complete theory of how the Earth works, from this beginning. One of the leading players in what was very much a team game was Dan McKenzie (born 1942), of the University of Cambridge, who recalls¹⁰ that it was a talk given by Hess in Cambridge in 1962, when McKenzie was still an undergraduate, that fired his imagination and set him thinking about the problems remaining to be solved by the model, and seeking other evidence to support it. Slightly more senior geophysicists in Cambridge were similarly inspired by that talk, and two of them, graduate student Frederick Vine (1939–1988) and his thesis supervisor Drummond Matthews (born 1931), combined the following year in a key piece of work linking the evidence for geomagnetic reversals with the sea-floor spreading model of continental drift.

By the early 1960s, as well as the growing mass of data about the magnetic history of the Earth obtained from the continents, the pattern of magnetism over parts of the sea bed had begun to be mapped, using magnetometers towed by survey ships. One of the first of these detailed surveys was carried out in the northeast Pacific, off Vancouver Island, around a geological feature known as the Juan de Fuca Ridge. Such surveys had shown a stripy pattern of magnetism in the rocks of the sea bed, with the stripes running more or less north–south; in one stripe the rocks would be magnetized in line with the present-day geomagnetic field, but in adjacent stripes the rocks would have the opposite magnetism. When plotted on a map, with one orientation shaded black and the other in white, the pattern resembles a slightly distorted bar code. Vine and Matthews suggested that these patterns were produced as a result of sea-floor spreading. Molten rock flowing from an oceanic ridge and setting would be magnetized with the magnetism corresponding to the Earth’s field at the time. But the continental evidence showed that the Earth’s magnetic field reversed direction from time to time.¹¹ If Vine and Matthews were correct, it meant two things. First, the pattern of magnetic stripes on the ocean floor should be correlated with the pattern of geomagnetic reversals revealed by continental rocks, providing a way to check the two patterns against one another and refine the magnetic dating of rocks. Second, since, according to Hess, crust spread out evenly on both sides of an oceanic ridge, the pattern of magnetism seen on one side of such a ridge should be the mirror image of the pattern seen on the other side of the ridge. If so, this would be striking confirmation that the sea-floor spreading model was a good description of how the Earth worked.

With the limited data available in 1963, the arguments put forward by Vine and Matthews could only be suggestive, not conclusive evidence in support of sea-floor spreading and continental drift. But Vine, in collaboration with Hess and with the Canadian geophysicist Tuzo Wilson (1908–1993), developed the idea further, taking on board new magnetic data that were coming in from around the world, and soon made the case compelling. Among the key contributions Wilson made was the realization that a spreading ridge like the one running down the Atlantic Ocean need not be a continuous feature, but could be made up of narrow sections which got displaced sideways from one another (along so-called transform faults), as if they were not one wide conveyor belt but a series of narrow conveyor belts lying side by side; he also played a major part in packaging many of the ideas of the new version of continental drift into a coherent whole. He was a leading advocate for these ideas, and coined the term ‘plate’ for one of the rigid portions of the Earth’s crust (oceanic, continental or a combination of both) that are being moved around by the forces associated with sea-floor spreading and continental drift.

The clinching evidence in support of the sea-floor spreading model came in 1965, when a team on board the research vessel Eltanin carried out three transverse magnetic surveys across a ridge known as the East Pacific Rise. The surveys showed a striking similarity between the magnetic stripes associated with the East Pacific Rise and those of the Juan de Fuca Ridge, further to the north – but they also showed such a pronounced lateral symmetry, from the pattern on one side of the Rise to a mirror image of that pattern on the other side of the Rise, that when the charts were folded along the line denoting the ridge the two plots lay one on top of the other. The results were announced in April 1966 at a meeting of the American Geophysical Union, held in Washington, DC; they were then published in a landmark paper in the journal Science.¹²

Meanwhile, the traditional approach to gathering evidence in favour of continental drift had been getting a boost. In the early 1960s, Bullard (by now head of the Department of Geodesy and Geophysics in Cambridge) championed the case that geological evidence in support of the idea of drift had by then overcome the difficulties the model had encountered in the 1920s and 1930s, and he presented the case for drift to a meeting of the Geological Society in London in 1963. The next year, he helped to organize a two-day symposium on continental drift at the Royal Society, where all the latest work was discussed, but where, ironically, the greatest impact was made by a new version of a very old idea – a jigsaw-puzzle-like reconstruction of Pangea. This reconstruction used what was presented as an objective method, based on a mathematical rule for moving things about on the surface of a sphere (Euler’s theorem), and with the actual reconstruction to provide the ‘best fit’, defined mathematically, carried out by an electronic computer to provide an unbiased, objective matching. The result is strikingly similar to Wegener’s fit of the continents and, in truth, said little that was new. But in 1964 people were still impressed by electronic computers and, surely even more significantly, were also, as they had not been forty years previously, disposed by the gathering weight of other evidence to take continental drift seriously. Whatever the psychological reasons, the ‘Bullard fit’ of the continents, published in 1965,¹³ has gone down as a defining moment in the story of the development of the theory of continental drift.

By the end of 1966, the evidence in support of continental drift and sea-floor spreading was compelling, but had not yet been assembled into a complete package. Most of the Young Turks of geophysics tackled the problem, racing to be first to publish. The race was won by Dan McKenzie (fresh from the award of his PhD in 1966) and his colleague Robert Parker, who published a paper in Nature in 1967¹⁴ introducing the term plate tectonics for the overall package of ideas and using it to describe in detail the geophysical activity of the Pacific region – the Pacific plate, as it is now known – in terms of the way plates move on the surface of a sphere (Euler’s theorem again). Jason Morgan, of Princeton University, came up with a similar idea, published a few months later, and although many details remained to be filled in (and are still being tackled today), what is sometimes referred to as ‘the revolution in the Earth sciences’¹⁵ had been completed by the end of the year. The essence of plate tectonics is that seismically quiet regions of the globe are quiet because they form rigid plates (six large plates and about a dozen small ones, between them covering the entire surface of the globe). An individual plate may be made up of just oceanic crust or just continental crust, or both; but most of the interesting geological activity going on at the surface of the Earth happens at the boundaries between plates – plate margins. Constructive margins are regions where, as we have seen, new oceanic crust is being made at ocean ridges and spreading out on either side. Destructive margins are regions where one plate is being pushed under the edge of another, diving down at an angle of about 45 degrees and melting back into the magma below. And conservative margins are regions where crust is neither being created nor destroyed, but the plates are rubbing sideways past one another as they rotate, as is happening along the San Andreas Fault today. Evidence such as the existence of ancient mountain ranges and former sea beds in the hearts of continents today shows that all this tectonic activity has been going on since long before the breakup of Pangea, and that supercontinents have repeatedly been broken up and rebuilt in different patterns by the activity going on on the surface of the restless Earth.

When the Open University was founded in Britain in 1969, these ideas and the rest of the package that made up plate tectonics were already becoming familiar to the professionals, and had been reported in the pages of popular magazines such as Scientific American and New Scientist, but had not yet found their way into the textbooks. In order to be bang up to date, in keeping with the vibrant image of the young institution, the staff at the Open University rapidly put together their own text, the first built around the global theory of plate tectonics. Since we have to draw a line somewhere, the publication of Understanding the Earth¹⁶ in 1970, neatly at the end of the decade which saw the ‘revolution’ in the Earth sciences, can be conveniently (if somewhat arbitrarily) taken as the moment when continental drift became the new orthodoxy – the last great triumph of classical science.

The fact that the continents have drifted, once established, helped to provide a new basis for understanding other features of the Earth, and in particular the relationship between living things and the changing global environment. The value of the insight this provides can be highlighted by one example. Alfred Russel Wallace, during his time on the islands of the Malay Archipelago, noticed that there was a distinct difference between the species to the northwest and the species to the southeast. This region between Asia and Australia is almost completely filled with islands, ranging in size from Borneo and New Guinea down to tiny atolls, and at first sight provides no insuperable barrier to the movement of species in both directions. Yet Wallace found that you could mark a narrow band on the map (now known as the Wallace line), running more or less from southwest to northeast between Borneo and New Guinea, with a distinctive Asian fauna to the northwest of this transition zone and a distinctive Australian fauna to the southeast, with little blurring between the two. This was a great puzzle at the time, but can be explained within the context of plate tectonics, where modern studies reveal that during the breakup of the southern supercontinent of Gondwanaland, Indo-Asia broke away first and moved to the northwest, where natural selection applied evolutionary pressures different from those at work in Australia-Antarctica that had been left behind. In a later phase of tectonic activity, Australia-New Guinea broke away from Antarctica, and moved northwards rapidly (by the standards of continental drift), eventually catching up with Asia. The two continents have only recently come into close proximity once again, and there has not yet been time for species from either side to mingle significantly across the Wallace line. Wegener himself commented on this possibility (writing in the third edition of his book, in 1924, only sixty-five years after Darwin and Wallace published their theory of natural selection, and just eleven years after Wallace had died); but it took plate tectonic theory to prove the point.

Continental drift is relevant to many aspects of the evolution of life on Earth, and is especially relevant to our theme of how science has refined our understanding of the relationship between human beings and the Universe at large, and our continual displacement from centre stage by new discoveries. Like Wallace, Charles Darwin explained how evolution works, but before he did that he was a geologist, and he would surely have been intrigued and delighted to learn of the modern understanding of the way in which continental drift and climate have come together to shape our species. It begins with the story of Ice Ages.

Even before the beginning of the nineteenth century, there were people who thought that glaciation in Europe had been much more extensive in the past than it is today. The most obvious evidence of this is the presence of huge boulders dumped far from the strata where they belong, having been carried there by glaciers which have since melted back and retreated – so it is hardly surprising that one of the first people to draw attention to these so-called ‘erratics’ was a Swiss, Bernard Kuhn, in 1787. But it is rather more surprising that he had this idea even though he was a clergyman; the received wisdom of the time was that all such phenomena could be explained by the effects of the Biblical Flood, whatever mountain folk might think from the evidence of their everyday contact with the effects of glaciers. Almost everyone subscribed to the received wisdom, and supporters of glaciation as the explanation of erratics were very much in the minority for decades to come. They included James Hutton, who was convinced by the evidence he saw on a visit to the Jura mountains; the Norwegian Jens Esmark, writing in the 1820s; and the German Reinhard Bernhardi, who knew of Esmark’s work and published an article in 1832 suggesting that the polar ice cap had once extended as far south as central Germany. This was just a year before Charles Lyell came up with the idea that erratics had indeed been transported by ice, but not by glaciers – he suggested, in the third volume of Principles of Geology, that great boulders could have been carried embedded in icebergs or resting on ice rafts, which floated on the surface of the Great Flood. But the chain which eventually led to a proper model of Ice Ages began not with any of the great names of nineteenth-century science, but with a Swiss mountaineer, Jean-Pierre Perraudin.¹⁷

Perraudin observed, up in the now ice-free mountain valleys, the way in which hard rock surfaces that did not weather easily had been scarred by something pressing down strongly on them, and realized that the most likely explanation was that they had been gouged by rocks scraped over them by ancient glaciers. In 1815, he wrote about these ideas to Jean de Charpentier, as he then was, a mining engineer who was also a well-known naturalist with geological interests that extended beyond the strict requirements of his profession. He was born Johann von Charpentier, in 1786, in the German town of Freiburg, but moved to Switzerland in 1813 and adopted a French version of his name; he stayed there, at Bex in the valley of the Aar, for the rest of his life, dying in 1855. De Charpentier found the idea of erratics being transported by glaciers too extravagant to accept at the time, although he was equally unimpressed by the idea that they had been carried to their present locations by water. Undaunted, Perraudin continued to present the evidence to anyone who would listen and found a sympathetic audience in the form of Ignace Venetz, a highway engineer whose profession, like de Charpentier’s, encouraged a broad knowledge of geology. Venetz gradually became persuaded by the evidence, including piles of debris found several kilometres beyond the end of the Flesch glacier, which seemed to be terminal moraines (heaps of geological rubbish left at the ends of glaciers) from a time when the glacier extended further down the valley. In 1829, he presented the case for former glaciation to the annual meeting of the Swiss Society of Natural Sciences, where just about the only person he convinced was de Charpentier, an old acquaintance with whom he had already discussed some of these ideas. It was de Charpentier who then picked up the baton, gathering more evidence over the next five years and presenting an even more carefully argued case to the Society of Natural Sciences in 1834. This time, nobody at all seems to have been persuaded (perhaps partly because Lyell’s ice-rafting model seemed to solve some of the puzzles involved in explaining erratics in terms of the Great Flood). Indeed, one member of the audience, Louis Agassiz, was so irritated by the notion that, in the best scientific tradition, he set out to disprove it and stop people discussing this nonsense once and for all.

Agassiz (who was christened Jean Louis Rodolphe, but was always known as Louis) was a young man in a hurry. He was born at Motier-en-Vuly, in Switzerland, on 28 May 1807, and studied medicine at Zurich, Heidelberg and Munich before moving on to Paris, in 1831, where he was influenced by Georges Couvier (then nearing the end of his life). He had already turned his attention to paleontology and soon became the world’s leading expert on fossil fishes. In 1832, Agassiz returned to Switzerland, where he was appointed professor of natural history at a new college and natural history museum being established in Neuchâtel, the capital of the region where he had been brought up. This part of Switzerland had a curious double status at the time. From 1707 onwards, although French-speaking, it had been part of the domain of the King of Prussia (except for a brief Napoleonic interregnum). In 1815, Neuchâtel joined the Swiss Confederation, but the Prussian link was neither formally acknowledged nor formally revoked (one reason why Agassiz studied in Germany) and the new college was supported by Prussian funds. When he took up the post, Agassiz already knew de Charpentier (they had met when Agassiz was still a schoolboy in Lausanne), who he liked and respected, and had visited the older man for busmen’s holidays during which they had probed the geology of the region around Bex. De Charpentier tried to persuade Agassiz that there had been a great glaciation; Agassiz tried to find evidence that there had not.

After another summer geologizing with de Charpentier around Bex in 1836, Agassiz was completely convinced, and took up the cause with all the evangelical enthusiasm of a convert. On 24 June 1837 he stunned the learned members of the Swiss Society of Natural Sciences (meeting, as it happened, at Neuchâtel) by addressing them, in his role as President, not with an anticipated lecture on fossil fishes, but with a passionate presentation in support of the glacial model, in which he used the term Ice Age (Eizeit; Agassiz lifted the term from the botanist Karl Schimper, one of his friends and colleagues). This time the idea really made waves. Not that people were convinced, but because Agassiz’s enthusiasm, and his position as President, meant that it could not be ignored. He even dragged the reluctant members of the Society up into the mountains to see the evidence for themselves, pointing out the scars on the hard rocks left by glaciation (which some of those present still tried to explain away as produced by the wheels of passing carriages). His colleagues were unimpressed, but Agassiz went ahead anyway, determined to find compelling evidence in support of the Ice Age model. To this end, Agassiz set up a small observing station (basically a little hut) on the Aar glacier to measure the movement of the ice by driving stakes into it and noting how quickly they moved. To his surprise, on summer visits over the next three years he discovered that the ice moved even faster than he had anticipated and that it could indeed carry very large boulders along with it. Fired by these discoveries, in 1840 Agassiz published (privately, at Neuchâtel) his book (Études sur les Glaciers), Studies on Glaciers which thrust the Ice Age model firmly into the arena of public debate.

In fact, Agassiz went completely over the top. It’s very difficult not to sit up and take notice of (for or against) a scientist who argues that the entire planet had once been sheathed in ice, and who makes his case in this kind of language:

The development of these huge ice sheets must have led to the destruction of all organic life at the Earth’s surface. The ground of Europe, previously covered with tropical vegetation and inhabited by herds of giant elephants, enormous hippopotami, and gigantic carnivora became suddenly buried under a vast expanse of ice covering plains, lakes, seas and plateaus alike. The silence of death followed…springs dried up, streams ceased to flow, and sunrays rising over that frozen shore…were met only by the whistling of northern winds and the rumbling of the crevasses as they opened across the surface of that huge ocean of ice.

These wild exaggerations managed to annoy even de Charpentier, who published his own, more sober (and less entertaining) account of the Ice Age model in 1841. It also planted Agassiz’s version of Ice Ages firmly in the catastrophist camp, as that ‘suddenly’ highlights, reducing its chances of receiving a welcome from Lyell and his followers. But evidence continued to accumulate, and as it did so, the fact that there had been at least one great Ice Age could no longer be ignored; before too long even Lyell was convinced that the model could be shorn of its catastrophist trappings and made acceptable to uniformitarians.

Several years earlier, Agassiz had visited Britain to study collections of fossil fish, staying for part of the time in Oxford with William Buckland, Lyell’s old mentor (but still a confirmed catastrophist), with whom he became friends. A year after Agassiz had startled his colleagues with his talk at Neuchâtel, Buckland attended a scientific meeting in Freiburg, where he heard Agassiz expound his ideas, then travelled on with his wife to Neuchâtel to see the evidence for himself. He was intrigued, but not immediately convinced. In 1840, however, Agassiz made another trip to Britain to study fossil fish and took the opportunity to attend the annual meeting of the British Association for the Advancement of Science (that year held in Glasgow) to present his Ice Age model. Following the meeting, Agassiz joined Buckland and another geologist, Roderick Murchison (1792–1871), on a field trip through Scotland, where the evidence in support of the model finally persuaded Buckland that Agassiz was right. Agassiz then went on to Ireland, while Buckland paid a visit to Kinnordy, where Charles and Mary Lyell had gone to stay after the Glasgow meeting. Within days, taking advantage of the evidence of former glaciation in the immediate vicinity, he had convinced Lyell, and on 15 October 1840 wrote to Agassiz:

Lyell has adopted your theory in toto!! On my showing him a beautiful cluster of moraines within two miles of his father’s house, he instantly accepted it, as solving a host of difficulties that have all his life embarrassed him. And not these only, but similar moraines and detritus of moraines that cover half of the adjoining counties are explicable on your theory, and he has consented to my proposal that he should immediately lay them all down on a map of the county and describe them in a paper to be read the day after yours at the Geological Society.¹⁸

The conversion of Lyell was not quite as dramatic as it seems, since (as this passage shows) he had already been worrying about the origin of these geological features; he had also visited Sweden in 1834, where he cannot have failed to notice the evidence of glaciation, even if he did not interpret it that way immediately. Compared with a Great Flood (still the preferred alternative), glaciation was uniformitarian – after all, there are glaciers on Earth today.

Buckland was referring in that letter to a forthcoming meeting of the Geological Society in London, where Agassiz was already listed as a speaker. In the end, papers by Agassiz himself, Buckland and Lyell, all in support of the Ice Age model, were spread over two meetings of the Society, and were read on 18 November and 2 December. It would be another twenty years or more before the model became fully accepted, but for our purposes we can cite those meetings where such established geological luminaries as Buckland and Lyell started preaching the gospel as the moment when the Ice Age model came in from the cold. The next big question to be answered would be, what made the Earth colder during an Ice Age? But before we look at how that question was answered, we should take a brief look at what happened to Agassiz after 1840.

In 1833, Agassiz had married Cécile Braun, a girl he had met as a student in Heidelberg. The couple were initially very happy, and had a son (Alexander, born in 1835) and two younger daughters (Pauline and Ida). But by the middle of the 1840s their relationship had deteriorated, and in the spring of 1845 Cécile had left Switzerland to stay with her brother in Germany, taking her two young daughters but leaving the older son to complete the current phase of his education in Switzerland. About this time (and a contributory factor to the breakup of the marriage), Agassiz was in severe financial difficulties because of an unwise involvement in an unsuccessful publishing venture. It was against this background that, in 1846, he left Europe for what was supposed to be a year-long trip to the United States, seeing geological features of the New World with his own eyes and giving a series of lectures in Boston. He was delighted both by the abundant evidence of glaciation that he saw – some within walking distance of the docks at Halifax, Nova Scotia, where the ship stopped before moving on to Boston – and by the discovery that his ideas about Ice Ages had not only preceded him across the Atlantic but had been widely accepted by American geologists. American geologists were equally delighted with Agassiz, and decided they wanted to keep him. In 1847, a new chair was established at Harvard for his benefit, solving his financial problems as well as giving him a secure academic base. He became professor of zoology and geology and stayed there for the rest of his life, establishing the Museum of Comparative Zoology in 1859 (the year Darwin’s Origin was published). Agassiz was a major influence on the development of the way his subjects were taught in the United States, emphasizing the need for hands-on investigation of natural phenomena; he was also a popular lecturer who helped to spread interest in science outside the campuses. Nobody is perfect, though, and to the end of his life Agassiz did not accept the theory of natural selection.

The American move proved timely for political, as well as for personal, reasons. In 1848, the wave of revolutionary activity in Europe engulfed Neuchâtel and the links with Prussia were finally severed. The college (which had actually been elevated to the status of an Academy in 1838, with Agassiz as its first rector) lost its funding and closed. The turmoil across Europe encouraged many naturalists to cross the Atlantic, some coming to work with Agassiz and boosting the status of the work going on at Harvard. Also in 1848, news came from Europe that Cécile had died, of tuberculosis. Pauline and Ida went to stay with their Swiss grandmother, Rose Agassiz, while their brother (who had joined the family in Freiburg just a year before) stayed with his uncle in Freiburg to finish his schooling. In 1849, Alexander joined Louis in Cambridge, Massachusetts; he eventually became a distinguished naturalist and founded the American branch of the Agassiz family. In 1850, Louis got married for the second time, to Elizabeth Cary, and brought his two daughters, then aged 13 and 9, over from Europe to join the family. He enjoyed nearly a quarter of a century of both domestic happiness and academic success in his new country, and died in Cambridge on 14 December 1873.

The roots of what is sometimes called the astronomical theory of Ice Ages go back to Johannes Kepler’s discovery, early in the seventeenth century, that the orbits of the planets (including the Earth) around the Sun are elliptical, not circular. But the story really begins with the publication in 1842, soon after Agassiz had published his own book on Ice Ages, of a book called Révolutions de la mer (Revolutions of the Sea) by the French mathematician Joseph Adhémar (1797–1862). Because the Earth moves in an ellipse around the Sun, for part of its orbit (part of the year) it is closer to the Sun than it is at the other end of its orbit (the other half of the year). In addition, the axis of the spinning Earth is tilted relative to a line joining the spinning Earth with the Sun, at an angle of 23½ degrees from the vertical. Because of the gyroscopic effect of the Earth’s rotation, on a timescale of years or centuries, this tilt always points in the same direction relative to the stars, which means that as we go around the Sun, first one hemisphere and then the other is leaning towards the Sun and gets the full benefit of the Sun’s warmth. That is why we have seasons.¹⁹ On 4 July each year the Earth is at its furthest distance from the Sun and on 3 January it is at its closest – but the difference amounts to less than 3 per cent if its 150 million kilometres average distance from the Sun. The Earth is furthest from the Sun in northern hemisphere summer, and is therefore moving at its slowest in its orbit then (remember Kepler’s laws). Adhémar reasoned (correctly) that because the Earth is moving more slowly during southern winter, the number of hours of total darkness experienced at the South Pole in winter is longer than the number of hours of continual daylight that the same region receives in southern summer, when the Earth is at the opposite end of its orbit and moving fastest. He believed that this meant that the south polar region was getting colder as the centuries passed, and that the Antarctic ice cap (which he believed to be still growing) was proof of this.

But the same thing can happen in reverse. Like a spinning top, the Earth wobbles as it rotates, but, being much bigger than a child’s top, the wobble (known as the precession of the equinoxes) is slow and stately. It causes the direction in which the axis of rotation of the Earth points relative to the stars to describe a circle on the sky once every 22,000 years.²⁰ So 11,000 years ago, the pattern of the seasons relative to the elliptical orbit was reversed – northern winters occurred when the Earth was furthest from the Sun and moving most slowly. Adhémar envisaged an alternating cycle of Ice Ages, with first the southern hemisphere and then the northern hemisphere, 11,000 years later, being covered by ice. At the end of an Ice Age, as the frozen hemisphere warmed up, he imagined the seas gnawing away at the base of a huge ice cap, eating it into an unstable mushroom shape, until the whole remaining mass collapsed into the ocean and sent a huge wave rushing into the opposite hemisphere – which is where the title of his book came from. In fact, the entire basis of Adhémar’s model was as shaky as he imagined those collapsing ice sheets to be. The idea that one hemisphere of the Earth is getting warmer while the other is getting colder is just plain wrong. As the German scientist Alexander von Humboldt (1769–1859) pointed out in 1852, astronomical calculations dating back more than a hundred years to the work of the French mathematician Jean d’Alembert (1717–1783) showed that the cooling effect Adhémar relied on is exactly balanced (it has to be exact, because both effects depend on the inverse square law) by the extra heat that the same hemisphere receives in summer, when the Earth is closest to the Sun. The total amount of heat received by each hemisphere during the course of a year is always the same as the total amount of heat received over the year by the opposite hemisphere. In the twentieth century, of course, as the understanding of the geological record improved and radioactive dating techniques became available, it became clear that there is no pattern of alternating southern and northern glaciations 11,000 years apart. But although Adhémar’s model was wrong, his book was the trigger which set the next person in the story thinking about orbital influences on climate.

James Croll was born in Cargill, Scotland, on 2 January 1821. The family owned a tiny piece of land, but their main source of income was from the work of Croll’s father as a stonemason. This meant that he travelled for much of the time, leaving his family to cope with the farming. The boy received only a basic education, but read avidly and learned the basics of science from books. He tried working at various trades, starting out as a millwright, but found that ‘the strong natural tendency of my mind towards abstract thinking somehow unsuited me for the practical details of daily work’.²¹ The situation was complicated further when his left elbow, injured in a boyhood accident, stiffened almost to the point of uselessness. This restricted Croll’s opportunities for work, but gave him even more time to think and read. He wrote a book, The Philosophy of Theism, which was published in London in 1857 and, amazingly, made a small profit. Two years later, he found his niche with a job as janitor at the Andersonian College and Museum in Glasgow. ‘Taking it all in all,’ he wrote, ‘I have never been in a place so congenial to me…My salary was small, it is true, little more than sufficient to enable me to subsist; but this was compensated by advantages for me of another kind.’ He meant access to the excellent scientific library of the Andersonian, peace and quiet, and lots of time to think. One of the things Croll read there was Adhémar’s book; one of the things he thought about was the way changes in the shape of the Earth’s orbit might affect the climate.

This idea built from the detailed analysis of the way the Earth’s orbit changes with time, which had been carried out by the French mathematician Urbain Leverrier (1811–1877). Leverrier is best remembered for his work that led to the discovery of the planet Neptune in 1846 (the same calculations were made independently in England by John Couch Adams (1819–1892)). This was a profound piece of work which predicted the presence of Neptune on the basis of Newton’s laws and the way in which the orbits of other planets were being perturbed by an unseen gravitational influence, after allowance had been made for the gravitational influence of all the known planets on each other. It was far more profound than the discovery of Uranus by William Herschel (1738–1822) in 1781, even though that caused much popular excitement as the first planet discovered since the time of the Ancients. Herschel’s discovery was lucky (in so far as building the best telescope in the world and being a superb observer was lucky). Neptune’s existence was (like the return of Halley’s comet in 1758) predicted mathematically, a great vindication of Newton’s laws and the scientific method. But the prediction involved horrendously laborious calculations with paper and pencil in those pre-computer days, and one of the fruits of those labours was the most accurate analysis yet of how the shape of the Earth’s orbit changes, on a timescale roughly 100,000 years long. Sometimes, the orbit is more elliptical and sometimes it is more circular. Although the total amount of heat received by the whole planet over an entire year is always the same, when the orbit is circular, the amount of heat received by the planet from the Sun each week is the same throughout the year; when the orbit is elliptical, more heat is received in a week when it is close to the Sun than in a week when it is at the other end of its orbit. Could this, Croll wondered, explain Ice Ages?

The model he developed assumed that an Ice Age would occur in whichever hemisphere suffered very severe winters, and he combined both the changes in ellipticity calculated by Leverrier and the effect of the precession of the equinoxes to produce a model in which alternating Ice Ages in each hemisphere are embedded in an Ice Epoch hundreds of thousands of years long. According to this model, the Earth had been in an Ice Epoch from about 250,000 years ago until about 80,000 years ago, since when it had been in a warm period between Ice Epochs, dubbed an Interglacial. Croll went into much more detail, including a sound discussion of the role of ocean currents in climate, in a series of papers which began with his first publication on Ice Ages in the Philosophical Magazine in 1864, at the age of 43. His work immediately attracted considerable attention, and Croll soon realized his lifelong ambition of becoming a full-time scientist. In 1867, he accepted a post with the Geological Survey of Scotland, and in 1876, the year after the publication of his book Climate and Time, he was elected as a Fellow of the Royal Society (possibly the only former janitor to receive this honour). Another book, Climate and Cosmology, followed in 1885, when Croll was 64. He died in Perth on 15 December 1890, having seen his Ice Age model become widely accepted and influential, even though, in fact, there was very little hard geological evidence to back it up.

In Climate and Time, Croll had pointed the way for further improvements to the astronomical model of Ice Ages by suggesting that changes in the tilt of the Earth might also play a part. This is the tilt, now 23½ degrees, that causes the seasons. In Croll’s day, it was known that the tilt changes (with the Earth nodding up and down, between extremes of tilt of about 22 degrees and 25 degrees from the vertical), but nobody, not even Leverrier, had calculated precisely how much it nods and over what timescale (it actually takes about 40,000 years to nod from its most vertical all the way down and back up to where it started). Croll speculated that when the Earth was more upright an Ice Age would be more likely, since both polar regions would be getting less heat from the Sun – but this was no more than a guess. By the end of the nineteenth century, however, the whole package of ideas had begun to fall into disfavour, as geological evidence began to pile up indicating that the latest Ice Age had ended not 80,000 years ago, but between about 10,000 and 15,000 years ago, completely out of step with Croll’s hypothesis. Instead of warming 80,000 years ago, the northern hemisphere was then plunging into the coldest period of the latest Ice Age, just the opposite of what Croll’s model required (and an important clue, which nobody picked up on at the time). At the same time, meteorologists calculated that the changes in the amount of solar heating produced by these astronomical effects, although real, were too small to explain the great temperature differences between Interglacials and Ice Ages. But the geological evidence did by then show that there had been a succession of Ice Ages, and if nothing else, the astronomical model did predict a repeating rhythm of cycles of ice. The person who took up the daunting task of improving the astronomical calculations and seeing if the cycles did fit the geological patterns was a Serbian engineer, Milutin Milankovitch, who was born in Dalj on 28 May 1879 (making him just a couple of months younger than Albert Einstein).

In those days, Serbia had only recently (1882) become an independent kingdom after centuries of outside domination (mainly by the Turks), although it had been an autonomous principality under Turkish suzerainty since 1829, part of a ferment of Balkan states gaining increasing independence in the gap between the crumbling Turkish Empire to the south and the scarcely healthier Austro-Hungarian Empire to the north. Milankovitch, unlike Croll, received a conventional education and graduated from the Institute of Technology in Vienna with a PhD in 1904. He stayed in Vienna and worked as an engineer (on the design of large concrete structures) for five years before becoming professor of applied mathematics at the University of Belgrade, back in Serbia, in 1909. This was very much a provincial backwater compared with the bright lights of Vienna, where Milankovitch could have carved out a good career; but he wanted to help his native country, which needed more trained engineers, and thought he could do this best as a teacher. He taught mechanics, of course, but also theoretical physics and astronomy. Somewhere along the line, he also picked up his grand obsession with climate. Much later,²² he romantically dated the moment when he decided to develop a mathematical model to describe the changing climates of Earth, Venus and Mars to a drunken conversation over dinner in 1911, when Milankovitch was 32, although this can perhaps be taken with a pinch of salt. What matters is that about that time, Milankovitch did indeed set out on a project to calculate not only what the present-day temperatures ought to be at different latitudes on each of these three planets today (providing a way to test the astronomical model by comparison with observations, at least on Earth), but also how these temperatures had changed as a result of the changing astronomical rhythms – actual temperatures, not just the more vague claim that at certain times of these cycles one hemisphere of the Earth was cooler than at other times in the cycles. And all of this, it cannot be overemphasized, with no mechanical aids – just brain power, pencil (or pen) and paper – and not just for one planet, but three! This went far beyond anything that Croll had even contemplated, and even though Milankovitch started out with an enormous bonus when he discovered that the German mathematician Ludwig Pilgrim had already (in 1904) calculated the way the three basic patterns of eccentricity, precession and tilt had changed over the past million years, it still took him three decades to complete the task.

Climate is determined by the distance of a planet from the Sun, the latitude of interest and the angle at which the Sun’s rays strike the surface at that latitude.²³ The calculations are straightforward in principle but incredibly tedious in practice, and became a major part of Milankovitch’s life, occupying him at home for part of every evening. The relevant books and papers even travelled with him on holidays with his wife and son. In 1912, the first of a series of Balkan wars broke out. Bulgaria, Serbia, Greece and Montenegro attacked the Turkish Empire, swiftly achieving victory and gaining territory. In 1913, in a squabble over the spoils, Bulgaria attacked her former allies and was defeated. And all of this turmoil in the Balkans contributed, of course, to the outbreak of the First World War in 1914, following the assassination of Franz Ferdinand by a Bosnian Serb at Sarajevo on 18 June that year. As an engineer, Milankovitch served in the Serbian army in the First Balkan War, but not in the front line, which gave him plenty of time to think about his calculations. He began to publish papers on his work, showing in particular that the tilt effect is even more important than Croll had suggested, but since these were published in Serbian at a time of political upheaval, little notice was taken of them. When the First World War broke out, Milankovitch was visiting his home town of Dalj, when it was overrun by the Austro-Hungarian army. He became a prisoner of war, but by the end of the year his status as a distinguished academic had earned him release from prison, and he was allowed to live in Budapest and work on his calculations for the next four years. The fruits of these labours, a mathematical description of the present-day climates of Earth, Venus and Mars, were published in a book in 1920 to widespread acclaim. The book also included mathematical evidence that the astronomical influences could alter the amount of heat arriving at different latitudes sufficiently to cause Ice Ages, although Milankovitch had not yet worked out the details. This aspect of the work was, however, immediately picked up by Wladimir Köppen, and led to a fruitful correspondence between Köppen and Milankovitch, and then to the incorporation of these ideas in Köppen’s book on climate with Alfred Wegener.

Köppen brought one key new idea to the understanding of how the astronomical rhythms affect climate on Earth. He realized that what matters is not the temperature in winter, but the temperature in summer. At high latitudes (he was thinking particularly in terms of the northern hemisphere), it is always cold enough for snow to fall in winter. What matters is how much of the snow stays unmelted in summer. So the key to Ice Ages is cool summers, not extra-cold winters, even though those cool summers go hand in hand with relatively mild winters. This is exactly the opposite of what Croll had thought, immediately explaining why the latest Ice Age was intense about 80,000 years ago and ended about 10,000–15,000 years ago. When Milankovitch took detailed account of this effect, calculating temperature variations on Earth at three latitudes (55 degrees, 60 degrees and 65 degrees North) he got what seemed to be a very good match between the astronomical rhythms and the pattern of past Ice Ages as indicated by the geological evidence available in the 1920s.

With the publicity given to these ideas by Köppen and Wegener in their book Climates of the Geological Past, for a time it seemed that the astronomical model of Ice Ages had graduated into a full-blown theory. In 1930, Milankovitch published the results of even more calculations, this time for eight different latitudes, before moving on to calculate, over the next eight years, just how the ice sheets would respond to these changes in temperature. A book summing up his life’s work, Canon of Insolation and the Ice Age Problem, was in the press when German forces invaded Yugoslavia (which had been established following the First World War and incorporated Serbia) in 1941. At the age of 63, Milankovitch decided that he would fill in time during the occupation by writing his memoirs, which were eventually published by the Serbian Academy of Sciences in 1952. After a quiet retirement, Milankovitch died on 12 December 1958. But by then, his model had fallen out of favour because new and more detailed (though still very incomplete) geological evidence no longer seemed to match it as well as the older, even less accurate evidence had.

In truth, the geological data simply were not good enough for any conclusive comparison with the now highly detailed astronomical model, and whether or not a particular set of data matched the model did not reveal any deep truth about the way the world works. As with the idea of continental drift, the true test would only come from much better measurements of the geological record, involving new techniques and technologies. These culminated in the 1970s, by which time the astronomical model itself (now often called the Milankovitch model) had been improved to an accuracy he can never have hoped for, using electronic computers. The key geological evidence came from cores of sediment extracted from the sea bed, where layers of sediment have been laid down year by year, one on top of the other. These sediments can be dated, using now-standard techniques including radioactive and geomagnetic dating, and are found to contain traces of tiny creatures that lived and died in the oceans long ago. Those traces come in the form of the chalky shells left behind when these creatures die. At one level, the shells reveal which species of these creatures flourished at different times, and that in itself is a guide to climate; at another level, analysis of isotopes of oxygen from these shells can give a direct indication of the temperature at the time those creatures were active, because different isotopes of oxygen are taken up by living creatures in different proportions according to the temperature and how much water is locked away in ice sheets. All three astronomical rhythms clearly show up in these records as the pulsebeat of the changing climate over the past million years or more. The moment when the model finally became established is generally taken as the publication of a key paper summarizing the evidence in the journal Science in 1976,²⁴ exactly a hundred years after the publication of Climate and Time. But that left one intriguing question, which turns out to be crucially important for our own existence. Why is the Earth so sensitive to these admittedly small changes in the amount of sunshine reaching different latitudes?

The answer brings us back to continental drift. Taking a step back from the close-up on climatic change provided by the Milankovitch model, the now well-understood (and well-dated) geological record tells us that the natural state of the Earth throughout most of its long history has been completely ice-free (except, perhaps, for the tops of high mountains). As long as warm ocean currents can get to the polar regions, it doesn’t matter how little sunlight they receive, since the warm water prevents sea ice from forming. But occasionally, as the scars of ancient glaciations reveal, at intervals separated by hundreds of millions of years, one or other hemisphere is plunged into a period of cold lasting for several million years; we can call this an Ice Epoch, lifting the term used by Croll and giving it a similar meaning but longer timescale. For example, there was an Ice Epoch which lasted for about 20 million years back in the Permian Era; this Ice Epoch ended around 250 million years ago. The explanation for such an event is that from time to time continental drift carries a large land surface over or near one of the poles. This does two things. First, it cuts off (or at least impedes) the supply of warm water from lower latitudes, so in winter the affected region gets very cold indeed. Second, the continent provides a surface on which snow can fall, settle and accumulate to build up a great ice sheet. Antarctica today provides a classic example of this process at work, which produces an Ice Epoch that is only slightly affected by the astronomical rhythms.

After the Permian Ice Epoch ended (which happened because continental drift opened the way once again for warm water to reach the polar regions), the world enjoyed about 200 million years of warmth, the time during which the dinosaurs flourished. But a gradual cooling began to set in about 55 million years ago, and by 10 million years ago glaciers returned, first on the mountains of Alaska but soon afterwards on Antarctica, where the ice sheet grew so much that by five million years ago it was bigger than it is today. The fact that glaciers spread in both hemispheres at the same time is an important insight. While Antarctica covered the South Pole and glaciers built up there in the way we have just described, the north polar region also cooled and eventually froze, even though the Arctic Ocean, not land, covered the pole. The reason for this is that continental drift gradually closed off an almost complete ring of land around the Arctic Ocean, shutting out much of the warm water that would otherwise have kept it ice-free. Notably, the presence of Greenland today deflects the Gulf Stream eastwards, where it warms the British Isles and northwestern mainland Europe instead. A thin sheet of ice formed over the polar ocean, and from about three million years ago much more ice lay over the surrounding land. This situation, with a polar ocean surrounded by land on which snow can fall and settle, but where it will melt away in hot summers, turns out to be particularly sensitive to the astronomical rhythms. For the past five million years or so, the Earth has been in what may be a unique state for its entire history, with ice caps over both poles, produced by two distinctly different geographical arrangements of land and sea. This, and in particular the geography of the northern hemisphere, makes the planet sensitive to the astronomical rhythms, which show up strongly in the geological record from recent geological times.

Within the current Ice Epoch, the effect of this pulsebeat of climate is to produce a succession of full Ice Ages, each very roughly 100,000 years long, separated by warmer conditions like those of the present day, Interglacials some 10,000 years long. By this reckoning, the present Interglacial would be coming to an end naturally within a couple of thousand years – less time than the span of recorded history. But the future is beyond the scope of this book. There are also lesser ripples of climate change superimposed on this principal pattern by the combination of rhythms investigated by Milankovitch. The sequence of these Ice Ages, dated by the radioactive technique using isotopes of potassium and argon, set in a little more than 3.6 million years ago. At just that time, our ancestors were living in the Great Rift Valley of East Africa (itself a product of plate tectonic activity), where an ancestral form of hominid was giving rise to three modern forms, the chimpanzee, the gorilla and ourselves.²⁵ It is just at this time that the fossil record provides direct evidence of a hominid that walked upright, in the form both of footprints made in soft ground which then set hard (like those handprints of the stars on the sidewalk in Hollywood) and of fossil bones. Although nobody can be sure, without the aid of a time machine, exactly what it was that turned one East African hominid from three to four million years ago into Homo sapiens, it is easy to make a case that the pulsebeat of climate played a key role, and hard to escape the conclusion that it was at least partially responsible. In East Africa, what mattered was not so much the temperature fluctuations, which were so important at high latitudes, as the fact that during a full Ice Age the oceans are so chilled that there is less evaporation and therefore less rainfall, the Earth is drier and the forests retreat. This would have increased competition among woodland hominids (including our ancestors), with some being forced out of the woods and on to the plains. There, there was intense selection pressure on these individuals, and only the ones who adapted to the new way of life would survive. Had the situation continued unchanged indefinitely, they might well have died out in competition with better-adapted plains dwellers. But after a hundred thousand years or so, conditions eased and the descendants of the survivors of that winnowing process of natural selection had a chance to take advantage of the expanding forest, breed in safety from the plains predators and build up their numbers. Repeat that process ten or twenty times, and it is easy to see how a ratcheting effect would have selected for intelligence and adaptability as the key requirements for survival on the fringes of the forest – while back in the centre of the forest, the most successful hominid lines adapted ever more closely to life in the trees, and became chimps and gorillas.

The story is perhaps as plausible as the idea of continental drift was in Arthur Holmes’s day. But even if the details are incorrect, it is hard to see how the close match between the climatic pattern which set in between three and four million years ago and the development of human beings from woodland apes, which also set in between three and four million years ago, can be a coincidence. We owe our existence to the combination of continental drift, establishing rare conditions ideal for the astronomical cycles to affect the Earth’s climate, and those astronomical rhythms themselves. The package involves basic physics (as basic as an understanding of convection, which drives continental drift), Newtonian dynamics and gravity (which explains the astronomical cycles and makes them predictable), chemistry (in analysing the samples from the sea bed), electromagnetism (for geomagnetic dating), an understanding of species and the living world built from the work of people like Ray and Linnaeus, and the Darwin–Wallace theory of evolution by natural selection. It is an insight which both puts us in perspective as just one form of life on Earth, created by the same process of natural selection that has created all other species, and triumphantly crowns three centuries of ‘classical’ science that began with the work of Galileo Galilei and Isaac Newton. Follow that, you might think. But by the end of the twentieth century, much of science had not so much followed that, as gone beyond classical science, changed in ways that are alien to the very foundations of the Newtonian world view. It all began right at the end of the nineteenth century, with the quantum revolution²⁶ that completely altered the way physicists thought about the world on very small scales.

^1.By Friedrich Viewege, Brunswick; for the definitive English translation of the fourth edition, see Bibliography.

^2.The daughter of Russian-born meteorologist Wladimir Köppen (1846–1940), a friend and colleague of Wegener.

^3.It took so long to pin the age down accurately because although the principles of the technique were known by 1910, the technology required to make measurements of the required precision took several decades to develop. As ever, science needs technology to progress, as much as technology needs science.

^4.Quoted by Lewis.

^5.Figures from Le Grand.

^6.The term ‘spreading sea-floor theory’ appeared in a paper published in 1961, and this was soon adapted to the more snappy ‘sea-floor spreading’.

^7.The broad features of the internal structure of the Earth are now quite well known, because the Earth’s interior has been probed by studying seismic waves produced in earthquakes and by nuclear bombs during the Cold War era of underground testing; the details, alas, are among the many detailed aspects of modern science that we do not have room to discuss.

^8.Literally, building activity, from the Greek word for builder.

^9.There is some evidence that geographical reconstructions of ancient supercontinents fit better if the Earth has actually expanded by a very small amount since the breakup of Pangea. This is intriguing, but even if the evidence stands up, the effect is only one of detail, not a major factor in driving continental drift.

^10.Conversation with JG, circa 1967.

^11.It is still not known exactly why this occurs, but it is thought to be a result of the dynamo effect operating in the Earth’s fluid core dying away to nothing, then building up again in the opposite sense. Intriguingly, the Sun, which is thought to have a similar internal dynamo, undergoes a similar pattern of magnetic reversals, but much more rapidly and much more regularly, associated with the roughly 11-year long sunspot cycle.

^12.W. C. Pitman and J. P. Heirtzler, Science, volume 154, pp. 1164–71, 1966.

^13.Blackett, Bullard and Runcorn, A Symposium on Continental Drift.

^14.Volume 216, pp. 1276–80.

^15.Of course, it was not a revolution; we hope we have made clear the way in which the ideas evolved, with new models patiently building on new data in the usual way that science progresses. The idea of scientific revolutions is essentially a myth beloved by sociologists who have never worked at the scientific coal face.

^16.Edited by Gass, Smith and Wilson.

^17.Not a mountaineer in the modern sporting sense, but somebody who made his living on the mountains, in this case by hunting chamois.

^18.Quoted in Elizabeth Carey Agassiz.

^19.On timescales of thousands and tens of thousands of years, the tilt is affected by various wobbles, which we are coming on to shortly.

^20.Modern calculations reveal that the cycle actually varies from 23,000 to 26,000 years, on still longer timescales, as a result of gravitational interactions with other bodies in the Solar System.

^21.For Croll’s autobiographical sketch, see Irons. Other quotes from Croll are from the same source.

^22.See Durch ferne Welten und Zeiten; this is the principal (if probably biased) source for biographical information about Milankovitch.

^23.And by the composition of the atmosphere, which is where the greenhouse effect comes in; but for these calculations we assume that the atmosphere has had the same composition over the past few million years.

^24.J. D. Hays, J. Imbrie and N. J. Shackleton, ‘Variations in the earth’s orbit: pacemaker of the ice ages’, Science, volume 194, pp. 112–1132, 1976.

^25.The dating of the split of the human line from the other African apes comes from direct measurements of their DNA, which provides a kind of ‘molecular clock’. This was finally established in the 1990s, as told by John Gribbin and Jeremy Cherfas in The First Chimpanzee (Penguin, London, 2001).

^26.Perhaps the only scientific ‘revolution’ that actually justifies the use of the term!