Consider, for a moment, the end of the world. Ice, as Robert Frost put it, would suffice.

Twenty thousand years ago, when glaciers swelled to malignancy and began spreading south, ice conquered the Earth. It bulldozed forests, filled valleys, swallowed chunks of mountains. It destroyed plants and animals that couldn’t adapt.

As the ice claimed more and more land, the climate became colder and colder. That only made more ice.

Yes, the Earth was on thin ice-or under thick ice, to be precise. The sheet that smothered what is now New York City, for instance, was over one mile thick. All told, the ice buried one third of Earth’s total land surface. And it stuck around for over 80,000 years.

Today we know that ice has overpowered our planet many times. In fact, Earth has endured at least ten ice ages in the last million years. But why? And when can we next expect to be buried in ice?

Truth is, we’re not 100 percent sure. Jose Rial, professor of geophysics at Carolina, says that ice-age theorists can be divided, roughly, into two camps. The “external” camp believes the ebb and flow of glaciers can be predicted by examining certain astronomical variations. A second “internal” camp focuses on the complexities of the Earth’s own climate system.

Rial has the perspective of an outsider, which allows him to keep a foot in both camps. Amiable and lively, he usually works with seismic waves, ground motion, and the like. Subjects closer to the Earth, so to speak. “I am not a climatologist,” Rial says. “I am a wannabe climatologist.” And when it comes to paleoclimatology-the science dealing with climates of past ages-he says he doesn’t have a reputation to protect.

But his paper in the journal Science (July 23, 1999) may change all of that.

Though Rial considers himself an outsider, his buoyant enthusiasm for Milutin Milankovitch-father of one of the oldest and most intriguing ice-age theories-suggests otherwise. “I believe him; I believe that he was absolutely right,” Rial says. “I have his book in my office, and I read it especially when I am depressed,” he laughs.

Some would say Milankovitch, a Serbian mathematician and engineer, had his head in the stars.

In 1920, a young Milankovitch resolved to develop a mathematical climate theory. Such a theory, he felt, should be based on a planet’s orbital variations. It would allow him to describe climates in the faraway reaches of the solar system-and to then reconstruct them far into the past. He would melt the ice-age question with math.

Milankovitch worked relentlessly. Using mathematically determined variations in Earth’s orbit, he painstakingly created timetables of the radiation Earth had received from the sun. By 1930, timetables in hand, Milankovitch proposed that changes in the Earth’s orbit could bring on the ice. Evidence from reconstructed glacial history appeared to back him up, and he won a lot of converts.

But not all scientists were convinced. In fact, after nearly 70 years, the debate is still lively.

Like Milankovitch, Rial likes to think big. Lately he’s been studying the Earth’s orbital variations. Earth’s solar orbit constantly expands and contracts, oscillating in shape between an almost perfect circle and an ellipse. This continual cosmic waltz is known as the Earth’s eccentricity. The whole cycle-from greatest to least eccentricity, and back-takes roughly 95,000 years.

But eccentricity has a second component-as Earth’s orbit shrinks and stretches, it also rotates. A full cycle lasts 413,000 years (see illustration).

These variations make the Earth’s orbit complex, but not so complex that we can’t understand it. “Astronomers can calculate every little variation with exquisite precision,” Rial says.

According to the Milankovitch camp, Earth’s orbital variations act as a celestial pacemaker to establish and drive climate rhythms. Meaning that if Milankovitch was right, the ice ages should keep time.

But they don’t. Earth’s interglacials-the time periods between ice ages-have been disturbingly erratic. Some have lasted only 80,000 years; others have lingered as long as 120,000.

Scientists can reconstruct climatic conditions of the past by drilling sediment cores from the ocean floor. A core’s concentration of heavy oxygen isotope-which has two more neutrons than common or “light” oxygen-hints at the history of ice on Earth.

Water containing heavy oxygen doesn’t readily evaporate. Rain, snow and ice-being primarily evaporated ocean water-are relatively high in light oxygen. When snow and ice don’t melt, the oceans tend to become richer in heavy oxygen. So if a sediment layer has a high heavy oxygen concentration, it was formed when the Earth was host to a lot of ice.

If ice growth is influenced by Milankovitch’s orbital frequencies, then oceanic sediment data should bear him out. And indeed, some heavy oxygen signals-which typically show up as multiple peaks in the sediment data-do peak at frequencies related to those predicted by Milankovitch. But other heavy oxygen signals have frequencies Milankovitch didn’t predict.

What’s more, one of Milankovitch’s frequencies has gone missing: the 413,000-year eccentricity signal has been notoriously absent from sediment data. And that’s given a lot of scientists a lot of headaches.

It has also damaged the astronomical theory’s credibility. Some scientists have looked away from Milankovitch, focusing instead on the way Earth tends to dip and rise within its orbital plane. That movement-known as orbital inclination-could, according to the logic, bring the Earth into the path of a giant dust ring. Sunlight would be filtered, the atmosphere would cool, glaciers would spread.

Rial finds this dust-ring scenario unlikely. The theory was sparked by a very strong heavy oxygen signal that had appeared in sediments dating back 600,000 years. But to Rial, the signal’s strength was suspicious.

He doubted that a mere 600,000 years’ data could accurately reflect the influence of orbital eccentricity. Such a short time window, Rial felt, would compress what might be several signals into one artificially strong peak. So he reanalyzed the “dust-ring” signal using a time window of one million years. Where other scientists had seen one strong peak, Rial found three.

Rial turned his attention to the interglacials. Though their lengths have varied, he noticed that they completed a full cycle-from longer to shorter durations and back-about every 400,000 years. Were the interglacials somehow reacting to the missing 413,000-year eccentricity signal?

Rial decided to find out. His inspiration? The radio dial.

In frequency modulation-the “FM” in FM radio-one wave modulates, or varies the frequency of, a second wave. An FM station’s frequency-say, 101.5-is called a “carrier” wave. The carrier wave is modulated by a broadcast wave-a music or voice wave, for example-which changes the carrier’s frequency. The resulting wave has a wider frequency band, including frequencies slightly higher and lower than 101.5.

Rial did a little frequency-modulating of his own. Using a computer simulation, he imposed the 413,000-year eccentricity cycle onto the 95,000-year eccentricity cycle, in effect making the 413,000-year cycle a modulator. Then he compared existing sediment records to the data produced by his simulation.

They matched almost perfectly. Rial then understood why the interglacials have seemed so unpredictable: “That 413,000-year period is transforming the 95,000-year period into a little more than 95,000, then a little less than 95,000,” he says. As that signal oscillates, so does interglacial length.

And the missing 413,000-year eccentricity signal? FM theory can explain that, too. In frequency modulation, the modulator-in this case, the 413,000-year signal-doesn’t show up in the final signal. Rather, it widens the signal’s frequency by creating sidebands, or peaks distributed symmetrically on both sides of the carrier.

That, Rial says, is why heavy oxygen signals often show multiple peaks-and why some of those peaks are not directly related to Milankovitch frequencies. Think back to the strong 100,000-year “dust-ring” signal. When Rial discovered that it was actually made up of three distinct signals, he was seeing sidebands.

Rial tested his idea by predicting where the sidebands for other signals should be. “Once you’ve caught the idea that it is frequency modulation,” he says, “then you can predict the sidebands, and boom! the sidebands are there.”

He later found evidence suggesting that the 413,000-year signal also modulates other orbital signals. “The modulator is hidden, but it’s moving the whole thing,” he says. “People like this research because it makes sense; it explains what were considered disturbing features of the data. And, it keeps the Milankovitch idea alive.”

So was Milankovitch right?

He was right, but not right all the way,” Rial says. From a certain perspective, Earth’s seemingly erratic interglacials actually occur in regular cycles.

But this regularity,” Rial says, “is produced not by the astronomical signal, but by the climate system itself, acting on-and distorting-the astronomical signal. If my model stands the test of time, it will strongly support Milankovitch. But it will also give its due to the climate system itself, which is capable of transforming the Milankovitch signal into something barely recognizable.”

And that’s where our notions of climate get a little foggier. “Earth’s climate system consists of the atmosphere, the ice caps, the oceans and waters in general, the biology-vegetation has a lot to do with climate-and even the land has an impact,” Rial says. Not to mention greenhouse gases, volcanoes, and humans. “The whole is greater than the sum of its parts,” Rial says. “How all those systems together respond to the astronomical influences is anybody’s guess at this point.”

We do know that the response is nonlinear. “Earth’s climate system doesn’t obey the usual mathematics that we are so accustomed to using,” Rial says.

Instead, the climate system resists the astronomical theory’s neatness. “For instance,” Rial says, “imagine if the heat from the sun were doubled: does the temperature on Earth double? Not necessarily-the temperature may triple, or may increase by four times, or may even decrease. The climate system is very much like a living system.”

Rial is awed by the interdisciplinary implications of what he has learned by reading articles in such fields as physiology-frequency modulation, as it happens, also occurs in the brain’s neuronal communications. “Mathematically, the systems are so similar that it’s almost unbelievable-but it’s really just the same language that nature uses to do these things,” he says.

Rial would like to use that language to develop a model for climate. So he’s teamed up with Cheri Anaclerio, a graduate student in biomedical engineering. “There might be some pattern behind the seemingly random jumps that the climate system makes,” Rial says. “We’d like to find that pattern and create a model-that’s where we’re going.”

If Rial and Anaclerio can come up with such a model, they might be able to see into the future-climate-wise, at least. “We’re after what’s going to happen in the next fifty years,” Rial says. “But we’re starting from very far back, taking the long view. We’re trying to let nature tell us where to go.”

But for now, we can all breathe easy. Rial predicts another 60,000 years will pass before an ice age claims the Earth again.

Illustration: Orbital eccentricity and sidebands