There are few things in life more exhilarating than getting a really good idea — one that just sings as it solves a problem that's been bothering you for God knows how long. We've all had this experience. Sometimes the ideas work out, and many illustrious careers have been built on such flashes of insight. Sometimes the ideas are total flops, in which case we bury them and move on.
The sciences are a branch of human endeavor in which ideas are the main item of commerce, the principle coin of the realm. Over the centuries, the scientific community has developed a complex set of rules about how ideas are to be evaluated, as well as some pretty definite criteria that tell you when they can be accepted. So if you want to follow the story of a good idea, what better place to look than science?
As it turns out, readers of this column already have an excellent example at hand. In the August 1993 Phenomena you met Douglas MacAyael of the University of Chicago, a young scientist who had just had a good idea about why Earth's climate developed as it did during the last ice age. Doug's idea has since gone through the scientific mill; it's time to go back and see what happened.
To refresh your memory, Doug works in glaciology. Having spent months camping on Antarctic glaciers and years trying to model their flow with computers, he knows about how ice behaves when it piles up. The problem he addressed had to do with a strange phenomenon people had found in cores drilled out of the ocean floor in the North Atlantic. Geologists were amazed to discover successive layers of rock debris and gravel that seemed out of place: the rocks in those layers appear identical to stuff you'd expect to find on land in northern Canada. Other evidence suggested that these layers, which formed every 7,000 to 12,000 years, marked periods of rapid climate change. Average temperatures climbed more than 10 degrees — the equivalent of moving the climate of Atlanta to Boston — in a few decades, followed a few thousand years later by an equally rapid return to normal. These sudden shifts in climate, accompanied by out-of-place rocks being dumped into the North Atlantic, were called "Heinrich events" after the German scientist who first discovered them.
Doug's idea was that you could understand both the origin of the rocks in the ocean bottom cores and the dramatic shift in the weather in terms of the behavior of the ice sheet that covered North America over much of the past 80,000 years. The depth of the sheet would increase as snow fell and compressed into ice, but when the ice lying on top of Hudson Bay reached to a height of about 10,000 feet, the soft rocks underneath would crumble and mix with meltwater, forming a slippery paste, and the whole thing would slough down Hudson Strait and eventually into the ocean, sending out an armada of icebergs, each with a load of crushed rocks frozen into its undersides. When the icebergs melted, the rocks were dumped (hence the strange layers). At the same time, the additional fresh water changed the patterns of ocean currents while the absence of two vertical miles of ice changed wind patterns (hence the change in climate).
"The point of the model," he told me recently, "was that everything depended on the internal behavior of the ice sheet. It was the ice driving Earth's climate, rather than vice versa." In a profession full of people who want to understand Earth's climate, a notion like this can be regarded as anything from a brilliant flash of insight to the rankest heresy. This is where the historic procedures of the scientific community come into play.
With an idea in hand, the first step is to submit a paper to a journal that in turn will vet it by a process known as peer review. The editors send the paper to one or more other scientists who evaluate it. The person submitting the paper is not told who the referees are, but normally they are chosen not just from his or her general field but from the writer's specialty. They advise the journal editor whether the paper is reasonable and worth publishing. Once the paper is published, it is subjected to a wider form of peer review. Any scientist anywhere in the world can submit a paper to the same journal calling attention to perceived inconsistencies, omissions or outright mistakes in the original paper.
In Doug's case, one of the anonymous referees realized that there was a big gap in his theory: he hadn't worked out how the sloughing ice could pick up the rocks and dump them into the ocean. A quick phone call initiated a collaboration that fixed that gap. This sort of criticism is a standard opening shot in any scientific debate, and in Doug's case it was well under way by the time his first paper was published.
The next (and more fundamental) critical step was not slow in starting. Call this the "there is another explanation for the whole thing" response. Often the most vociferous part of scientific debate, it usually consists of arguments between identifiable camps of individuals. Doug's "ice causes climate change" school was one such camp, which was opposed by a "climate change changes the ice" school, whose central idea was that it was global warming that caused the purging of the Hudson Bay ice sheet, rather than the growth of the ice sheet itself. This stage of scientific debate often resembles nothing so much as a trial in open court. It can be rancorous, it can be unsettling, but it is also necessary because, in the end, only those ideas that have been tested in this kind of fire can be trusted.
The debate is made more spirited by the very human tendency of the protagonists to believe that their particular theories explain everything there is to explain. "I was like a raccoon caught in the headlights," Doug now says with a rueful grin. "I was so obsessed with the power of the idea of ice sheet dynamics that I wanted it to be the whole picture and not just part of a larger whole. The idea was beautiful, simple and, in the end, wrong."
The "beautiful" and "simple" are understandable. What an easy world it would be if climate had been governed by the rise and fall of a single glacier. To get to the "wrong," though, you have to understand one more crucial weapon in the testing of a new idea: new data. When snow falls and packs down into ice, it carries all sorts of information with it, and that information gets locked into the glacier. The relative abundances of different versions of oxygen (what scientists call the "isotope ratio") in the ice molecules tell us what the temperature was when the snow fell.
What makes this so useful is that each year's snowfall makes an easily identified line in the ice. This means that as we drill down, these lines act like tree rings and allow us to keep track of the year. When two drilling teams (one European and one American) hit the rock 10,000 feet under the surface of the Greenland ice sheet last year, they had cores with records of the past 200,000 years of North Atlantic climate.
And what a story they tell! The Heinrich events are there: every 7,000 to 12,000 years the world had experienced a cold wave — at the same times that rock debris was being dumped on the ocean floor. Immediately thereafter, temperatures would climb to above normal. The higher temperatures would last for several thousand years, and then the climate would return to normal.
The Greenland ice cores had more to say, however. Between the Heinrich events are a series of shorter cold snaps, which begin and end with temperature changes as rapid as those that mark the Heinrich events. Referred to informally as "flickers," these events start as small drops in temperature, but each succeeding one is colder than the last.
Looking closely at deep-ocean cores, scientists found a second series of sediment layers corresponding to the onset of each flicker. Sand in these layers featured a reddish iron coating that could be traced not to Hudson Bay but rather to the St. Lawrence Valley, which was also covered by ice at the time.
"So now we had two ice sheets going through the buildup-and-purge cycle," says MacAyael. Hudson Bay produced the Heinrich events, and the St. Lawrence caused the flickers. "The Saint [Lawrence] is farther south where the ice sheet got more snow, and so it built up more quickly. This may be why the flickers occur more often than the Heinrich events." (More snow means the St. Lawrence ice sheet reached the critical depth to make a slurry for fast flow more often than its counterpart in Hudson Bay.) So although the addition of a second ice sheet emptying into the Atlantic makes the picture more complicated, it is still consistent with the original grand hypothesis that climate changes are caused by ice dynamics.
More new information, however, changed everything. It turns out that the flickers build up in intensity between Heinrich events, but that the warming after the last (and coldest) flicker seems to coincide with the warming of the Heinrich event itself. The picture, then, is that the St. Lawrence ice sheet built up and purged several times as the ice dome over Hudson Bay was growing, but that when the ice over Hudson Bay let go, the ice over the St. Lawrence did the same. There's no reason why two ice sheets, each with different snowfall and structure, should be synchronized like that if climate depends only on the internal dynamics of the ice.
So the world turns out to be a lot more complicated than any of the different camps in the debate thought it would be. It's not "just the ice" or "just the climate," but some complicated mix of the two that determines what happens on our planet. At this writing, Doug and one of his graduate students, Christina Hulbe, are trying to see whether the effect of the climate might be to create a floating ice shelf off the Canadian coast (similar to the Ross ice shelf in Antarctica) that would, in effect, "put a cork in the bottle" and prevent the glaciers from discharging until the cork popped. Other people have other ideas, and the debate goes on. The effect of the original good idea, then, has been to drive the scientific community to a more realistic, and more complicated, view of the world's climate. Be it ever thus.
by James Trefil