Looking at the rhythmic curves of past cycles, one could hardly resist the temptation to extrapolate into the future. By the late 1980s, most calculations had converged on the familiar prediction that the natural Milankovitch cycle should bring a mild but steady cooling over the next few thousand years. As climate models and studies of past ice ages improved, however, worries about a swift descent into the next great glaciation — what many in the 1970s had tentatively expected — died away. Improved calculations said that the next ice age would not come naturally within the next ten thousand years or so. The conclusion was backed up in 2004 by a heroic new ice core from Antarctica that brought up data spanning the past eight glacial cycles.(53*)
The scientists who published these calculations always added a caveat. In the Antarctic record, atmospheric CO2 levels over the past 750,000 years had cycled between about 180 and 280 parts per million. The level in the late 20th century had now climbed above 370 and kept climbing. (The other main greenhouse gas, methane, was soaring even farther above any level seen in the long ice record.) Greenhouse warming and other human influences seemed strong enough to overwhelm any natural trend. We might not only cancel the next ice age, but launch our planet into an altogether new climate regime. The ice cores themselves gave convincing evidence of the threat, according to analyses published in the early 1990s. The “climate sensitivity” — the response of temperature to changes in carbon dioxide — could be measured for the last glacial maximum. The answer was in the same range that computer models were predicting for our future, raising confidence that the models were not far wrong.(53a)
In climate science, where everything is subtle and complex, it is rare for an issue to be settled. By the late 1980s, it did seem to be an established fact that ice ages were timed by orbital variations. The chief question that remained in the minds of most scientists was what kind of feedbacks amplified the effect. Yet some people challenged whether any of this was really understood. The feedbacks that helped drive glacial cycles remained uncertain. The cycles, most scientists now agreed, involved not only orbital variations in solar irradiation, but also a variety of geological effects. First came the massive settling and flow of continental ice sheets, but large-scale physical and chemical changes in the oceans might be important too. New evidence gave a particularly crucial role to changes in CO2 and other greenhouse gases. These changes were apparently driven not just by geochemistry and ocean circulation, but still more by changes in biological activity. And of course the biosphere depended in turn on climate — and not just temperature, but also trickier matters like fertilization of the seas by minerals eroded from glacial era deserts. Further peculiar influences were added to the list of possibilities almost every year.(54) It would take much more study to determine just what combination of effects determined the shape of glacial cycles.
In 1992, a more fundamental challenge was raised by the ingenious exploitation of a novel source of data: layers of calcite laid down in the desert oasis of Devils Hole, Nevada. The layers showed glacial and interglacial periods much like those seen in the ice cores. The dating (using uranium isotopes) failed to agree with Milankovitch calculations. The authors suggested that the timing of ice ages followed no regular cycle at all, but was driven wholly by “internal nonlinear feedbacks within the atmosphere-ice sheet-ocean system.”(55) A vigorous controversy followed, but in the end most climate scientists stuck by the Milankovitch theory. The Devils Hole measurements looked solid, but didn’t they represent only a strictly local effect?
Apparently the story (like most climate stories) was not simple. As two experts reviewing the problem put it, “climate is too complicated to be predicted by a single parameter.”(56) The faint variations of summer sunlight were effective only because the astronomical schedule somehow resonated with other factors — ice sheet and ocean dynamics, the bio-geochemical CO2 system, and who knew what else. The more precise the data got, the less precise seemed the match between Milankovitch and ice age cycles. Evidently when orbital effects served as a pacemaker, it was only by partially adjusting the timing of greater forces working through their own complex cycles. As one reviewer said, “The sheer number of explanations for the 100,000-year cycle… seems to have dulled the scientific community into a semipermanent state of wariness about accepting any particular explanation.”(57)
As researchers extracted more precise data from the distant past, they discovered that the weak 100,000-year orbital cycle had not always dominated the ice ages after all. Go back more than a million years, and it was the 40,000-year cycle that ruled. The reason for the switch was obscure. The grand puzzle of the ice ages stood unsolved — except insofar as scientists now understood that nobody would ever jump up with a neat single solution. There would instead be a long collective trudge through the intricacies of field data and models, gradually increasing our knowledge of all the interacting forces that drive climate cycles. The invaluable fruit of a century of ice ages research was the recognition of how complex and powerful all the feedbacks could be.
An important clue came from some especially good ice core records that showed a lag in the levels of CO2 and methane. They seemed to rise or fall a few centuries after a rise or fall in temperature. This confused many people, who thought the time lag contradicted the greenhouse theory of global warming. But in fact the lag was not good news. Scientists quickly realized that it strongly confirmed that the Milankovitch-cycle shifts in sunlight initiated a powerful feedback loop. Evidently the close of a glacial era came when a slight rise of temperature stimulated massive changes in gas levels, which drove the temperature still higher, which drove further changes in the gas levels, and so forth. Ice ages were thus the reverse of our current situation, where humanity was initiating the change by adding greenhouse gases. Once that began to warm the planet, would the feedbacks begin to drive things higher on their own?
It was now clear that not only the most obvious feedback, but also the most momentous one, was the connection between global temperature and greenhouse gas levels. Relatively straightforward analysis of the data showed that a doubled level of CO2 had always gone along with a rise of a few degrees in global temperature. It was a striking verification, with entirely independent methods and data, of what computer models had been predicting for the planet’s greenhouse future.
