Frozen Earth: The Once and Future Story of Ice Ages

by Doug Macdougall

  In his book, Milankovitch had calculated temperatures at various latitudes on Earth through the past 130,000 years. However, all the evidence then available indicated that the Pleistocene Ice Age with its multiple glacial advances and retreats stretched back much farther in time than this. Köppen asked if Milankovitch could extend the calculations to at least 600,000 years before the present so that they could make a detailed comparison between theory and the field evidence. It would not be necessary, he said, to do the calculations for every latitude. He suggested restricting attention to the swath between 55° and 65° north latitude, since that was the region from which much of the detailed evidence for glaciation had been gathered.

  Today, scientific results are frequently reported with the aid of elaborate, colorful graphs. Even in the 1920s, especially for interdisciplinary work, it was important to think about how the results would be presented. Milankovitch’s calculations would result in tables of numbers that showed the amount of heat received from the sun at various latitudes. Tables could convey the information, but they would not have the immediate impact of a visual representation. Köppen and Milankovitch discussed the problem at length, finally deciding that a graph showing the changes over time would be best. That would illustrate for the reader both the magnitude of the variations, and exactly when the coldest and warmest intervals had occurred. When he had finished his calculations, Milankovitch sent Köppen graphs showing the amount of solar energy received at 55, 60, and 65 degrees north latitude, plotted against time (figure 14). He had calculated these values at intervals of 10,000 years back to 650,000 years ago, and it had taken him just over three months of intense work. Instead of plotting the actual solar energy values, or the estimated temperatures, which he knew were problematic because of the redistribution of heat by the ocean and atmosphere, he plotted what he referred to as the “equivalent latitude.” This was an interesting idea because it showed immediately whether the latitudes he had chosen for his calculations received more (lower equivalent latitude) or less (higher equivalent latitude) solar radiation—usually referred to as “insolation”—than the same latitude does today. For example, Milankovitch’s work indicated that 230,000 years ago, during the northern summer, the Earth received only as much solar energy at 65° N latitude as it receives today ten degrees further north. The equivalent latitude was 75° N (figure 14), indicating that the summer temperatures were much cooler at that time.

  Köppen was pleased with the graphs. One of the reasons he had asked Milankovitch to calculate temperatures back to 600,000 years was that in the late 1800s, two German scientists, Albrecht Penck and Eduard Brückner, had developed a timescale for glaciation in the Alps that covered exactly that period. Based on observations of gravel terraces along Alpine river valleys, they had identified four glacial periods during this interval. Each terrace, they believed, corresponded to a period when active glacial erosion of the valley walls provided an abundant supply of gravel; when glaciation ended, the rivers could no longer transport much gravel and would begin to cut down through previously deposited sediments. As this was before the advent of radioactive dating techniques, they had to determine the age of each of the glacial periods by making assumptions about the rate at which the geological processes had operated. They simply measured the difference in elevation between successive terraces, and divided that distance by the rate at which they estimated the streams had eroded their beds. This was a very crude way to obtain the time that had elapsed between the formation of each of the terraces—and by implication, the successive glacial periods—but although most geologists recognized that the estimates were uncertain at best, real numbers published in papers by respectable scientists are seductive. Soon the four glacial periods described by Penck and Brückner became part of the vocabulary of European geologists, and the fact that American geologists had also mapped out drift deposits that seemed to come from four separate glacial advances gave their scheme added respectability. Even after other workers began to point out flaws in the Penck and Brückner field evidence, their version of the temperature history of Europe continued to be popular. It remained so into the 1940s before finally being discarded.

  The Penck and Brückner scenario was very much in vogue when Köppen first saw Milankovitch’s extended graph of equivalent latitudes. He realized at first glance that there was a close correspondence between the two—the times when Milankovitch’s graph indicated cold Northern Hemisphere summers corresponded well to the times of Alpine glaciation based on the gravel terraces. The coincidence between these independent records—one theoretical and the other derived from field evidence—was very convincing; it seemed almost too good to be true. Milankovitch published the graph in a paper under his own name, and Köppen and Wegener put it in their book on past climates, with due credit to Milankovitch. The Köppen and Wegener book, Die Klimate der geologischen Vorzeit (Climates of the Geological Past), published in 1924, reached a wide audience and quickly brought Milankovitch worldwide prominence. The similarity between his graphs and the Penck and Brückner glacial timescale was so persuasive that he was generally acknowledged to have proven the astronomical theory of climate.

  Milankovitch’s graph soon became a cornerstone of much ongoing ice age research. Rather than testing his results further against other estimates of glacial timing, many geologists instead simply accepted them and began to use the graphs as a way to date glacial deposits. This was precisely what James Croll, many years earlier, had predicted could be done with his own graphs of the Earth’s orbital variations.

  Milankovitch did not rest on his laurels when his work became more widely known. Instead, over the next decade he continued to refine his calculations. He pushed them back even farther into the past, to a million years before the present. He expanded the range so that his results extended from 5° to 75° latitude, both north and south of the equator. As more precise data on the masses of the planets and their orbits became available, he incorporated them into his calculations of the Earth’s orbital variations. Although these refinements did not materially change the conclusions that he, Köppen, and Wegener had drawn about past climate, they did provide a much more comprehensive picture of how the astronomical factors influence the solar energy budget over the Earth’s surface through time. In the midst of this work, Wegener, who by this time had become a close friend, died tragically during a scientific expedition on the Greenland ice cap. The news of his death reached Milankovitch in May 1931, and although he had been expecting the worst—Wegener had set out on a dogsled journey the previous September and had not been heard from in the intervening months—it was a great blow. Wegener had been only fifty, and there was much he had still wanted to accomplish. His body has never been found.

  Although Milankovitch’s calculations and the conclusion that he, Wegener, and Köppen had come to about the connection between glaciation and insolation changes gained widespread acceptance, there was one question that continued to nag him. Beyond the general idea that glaciation coincided with periods of cool northern summers, he had no concrete criterion for the initiation of a glacial period. Were the insolation variations themselves sufficient? Or was there some other factor involved?

  Milankovitch recognized, as had earlier workers, that the “snow line,” the elevation at which permanent snow exists in the mountains, is an important indicator of average temperature. The elevation of the snow line varies with latitude—at present, permanent snow and ice occur at sea level at the poles, while in the tropics, they exist only on the highest mountains. Milankovitch realized that if the astronomical variations affected average temperatures—as his calculations showed—then they would also be correlated with the rise and fall of the snow line at a particular latitude. The difficulty was to find a mathematical relationship connecting the elevation of the snow line with his results for insolation. He realized, as had Croll, that there is positive feedback involved, because snow is much more reflective than forest or open ground, and as snow cover increases,
more solar energy is reflected back into space. Fortunately, the appropriate measurements on snow and ice reflectivity had been made in the early 1930s, and Milankovitch could plug the relevant parameters into his equations with some confidence. The results were encouraging: the general pattern of his earlier curves remained, but the intensity of the cold periods increased when the reflective effects of the snow were included. When he calculated the altitude of the permanent snow line, he found that it descended to low elevations during the cold periods, and rose when it became warmer. In effect, a snow line near sea level meant that a permanent ice cap had formed. It appeared that the insolation variations, amplified by the effect of increasing snow cover, could indeed trigger the onset of glaciation.

  With the inclusion of snow reflectivity in his calculations, Milankovitch had completed work on all aspects of the problem that he thought were important. But he had published the results of his investigations in bits and pieces, in different languages and at different times. His most famous graph was known by most people because of its inclusion in Köppen and Wegener’s book. He was now corresponding with scientists from around the world, many of whom were working on the ice age problem, and most of them requested copies of his papers. There were no Xerox machines or electronic versions of his work, and he was quickly running out of the few copies his publishers had given him. He resolved to put all of his investigations of climate together in a book that would encompass everything, from the details of the equations necessary to calculate the motions of the planets around the sun to a discussion of ice ages. He called this massive work Canon of Insolation and the Ice Age Problem. It was written in German and published in 1941 (an English translation did not appear until 1969). Like Harlan Bretz, Milankovitch was very organized and systematic; his book begins with the classical laws of mechanics and then proceeds, step by step, through all of the calculations necessary to address the problem of climate variations.

  Canon of Insolation was written during the early part of World War II, and it took Milankovitch, he says, 539 days (ever the mathematician, Milankovitch is precise; he doesn’t say “about a year and a half,” but gives us the time to the day!). The last pages were printed in Belgrade in April 1941; a few days later, Germany invaded Yugoslavia. Belgrade was bombed and the printing shop where his book was being produced was reduced to rubble. Fortunately, however, most of the printed pages survived intact. Even under occupation, some things continued to work in Belgrade—Milankovitch’s rescued manuscript was bound and distributed, a brief happy occurrence in an otherwise unhappy time. His memoirs make it clear that his experience of life under German occupation was not pleasant. “Our civilized existence,” he wrote, “had disintegrated into a life of hard grind.”

  Figure 15.Milutin Milankovitch working at his desk in 1954. Photograph courtesy of his son, Vasko Milankovitch.

  Milankovitch was sixty-two and isolated from the world of science because of the war. The university was closed; his friend and colleague Köppen had died almost two years before, Wegener a decade earlier. His work on climate had been completed with the publication of Cannon of Insolation, and his scientific career was effectively over. Living conditions were grim, and the wait for the war’s end seemed interminable. Not content to be idle, Milankovitch decided to write a history of science—an activity, he said later, that kept him sane. It was published in Serbian at the end of the war. After the Germans had left, life in Yugoslavia under the communists was only marginally better. But Milankovitch had an international scientific reputation, and he was for the most part left in peace. He worked on his memoirs and some manuals for university courses (figure 15), and he was able to travel abroad occasionally. Late in 1958, at the age of seventy-nine, he died of complications from a stroke.

  By then, the astronomical theory of glaciation was not in good health, either. This idea, which had seemed so promising when James Croll first proposed it, was, for the second time, losing influence among scientists who had initially embraced it. Milankovitch’s work had revived the theory at a time when it had been all but forgotten, and his comprehensive calculations put it on a much firmer footing than Croll had been able to do. But once again issues of timing began to cast doubt on the theory. First, the gravel terraces of the Alps that had been thought to record glacial periods turned out to contain fossils incompatible with a cold climate. Later, it was discovered that the terraces had not been formed by glaciation, but had another origin altogether. The glacial timescale devised by Penck and Brückner, which had been one of the most important pieces of confirming evidence for Milankovitch’s theory, was completely meaningless. As if that were not enough, there was a second blow to the theory when a new dating method showed that glacial deposits in North America did not correspond well in age to predictions of the theory.

  The new technique was carbon-14 dating, invented in the late 1940s by Willard Libby and his students at the University of Chicago. It was an elegant method, making use of the fact that radioactive carbon-14 is continually produced in the Earth’s atmosphere by cosmic rays, but then begins to decay away when it is incorporated into organic material. It had potential applications in a wide range of subjects, and Libby was later awarded the Nobel Prize in chemistry for his work. For the geologists working on ice ages, the new technique was a godsend. Here, finally, was a method that could provide “absolute” dates for the deposits of the glacial periods. Unlike other methods commonly used for measuring ages in geology, the carbon-14 technique cannot be used to date rocks; it works only for organic material that was once alive and exchanging carbon with the atmosphere. This meant that a fossil in glacial drift, or a piece of wood preserved in a peat bog were perfect samples for dating. There was one problem, however: radioactive carbon-14 has a half-life that, in geological terms, is quite short. Especially in the early days of its application, it could only be used to date materials that were not more than a few tens of thousands of years old. In older deposits, so much of the radioactive carbon-14 had decayed away that the few remaining atoms could not be detected.

  Soon after Libby had demonstrated the feasibility of the carbon-14 method, others began to set up the necessary equipment, and before long, especially in the United States, there were a number of laboratories capable of making the analyses. Geologists studying ice ages had more than enough samples to keep them busy. Very quickly they established a timescale for the movement of the North American ice sheet by dating the moraines and drift that marked the margins of the ice at different times in its history. Even though they could not extend the analyses very far into the past, the carbon-14 data presented a far more complex story of glacial advances and retreats than Milankovitch’s graph, which showed single cold-summer spikes at 25,000 and 72,000 years ago. History was repeating itself. Just as Croll’s attempt to prove an astronomical cause for ice ages had foundered because the timing didn’t seem to be right, so too the carbon-14 analyses in North America, especially in light of the complete abandonment of the Penck and Brückner timescale for European glaciation, seemed to sound the death knell for Milankovitch’s revised and updated version of the theory. Furthermore, some meteorologists had looked again at the solar radiation balances calculated by Milankovitch and declared that the variations were just too small to produce drastic changes in climate, even when the effects of increased snow cover were incorporated. The pendulum was again swinging away from the astronomical theory. But the story has yet one more twist, one that brought Milankovitch’s calculations back to center stage.

  The evidence that revived the Croll-Milankovitch theory was discovered more than a decade after Milankovitch’s death, and it came from the depths of the ocean. Many geologists had recognized that sediments on the sea floor might provide a long-term record of climate and other environmental conditions on the Earth’s surface. Unlike moraines or loess or other glacial features on land, ocean sediments were presumed to accumulate slowly and continuously, century by century and millennium by millennium, without disturbance.
If it were possible to core into these sediments, it might be possible to retrieve a record of events far into the past. Indeed, James Croll had presciently suggested that ocean sediments might provide the best clues about glacial cycles.

  By the 1960s and 1970s, the technology for sampling the sea floor had improved to the point where it was possible to retrieve long cores of sediment reaching back millions of years into the past. Geochemists studying such cores noted that there are regular variations in the chemical composition of the sediments, and paleontologists examining fossils reported that there are similar alternations in the abundance of species that lived in warm and cold conditions. Oceanographers began to link these changes to glacial-interglacial cycles. But there was still the problem of assigning a timescale—accurate ages were needed to compare the dates of the sediments with the ages of glacial deposits on land and the timing of astronomical variations. Carbon-14 dating turned out to be more complicated for ocean sediments than for land deposits, and in addition there was the problem that it is limited to the past few tens of thousands of years. To probe farther into the past and investigate the cyclic changes that characterized the deep-sea sediment cores would require a new approach.

  The breakthrough that came to the rescue occurred in another, unrelated, area of earth science. Geophysicists studying the Earth’s magnetic field discovered that it has reversed periodically in the past—the north and south magnetic poles have switched positions. When rocks form on the Earth’s surface—when lava erupts from a Hawaiian volcano, for example—the magnetic minerals they contain line up their own small magnetic fields in the same direction as the Earth’s. By dating such rocks and measuring their magnetic orientation, a record has been built up of how the Earth’s magnetic field has varied in the past. Particularly important is the timing of the magnetic reversals, the geologically short intervals when the field switches from “normal” (today’s situation) to reversed, because these serve as markers or time lines that can then be used to date other events. Through the work of many different laboratories, the reversals have been dated quite accurately.