What Really Causes the Ice Ages?

The last million years on Earth have been one long ice age, interrupted regularly by interglacials, or brief periods of warmth. The warm spells have usually lasted between 10,000 and 20,000 years. We're in one now that began about 12,000 years ago. So any millennium now the temperature will drop, glaciers will grow, and ice sheets thousands of kilometers thick will advance on the continents, devouring a large fraction of the land on the planet.

For most of the last million years these cycles of glaciation have occurred worldwide every 100,000 years — 80,000 to 90,000 years of ice followed by 10,000 to 20,000 years of warmth. Why are they so regular?

In the mid-1800s, when the existence of ice ages was discovered, there was no way to know the cause or frequency of these climate changes. James Croll, a self-taught Scottish scientist, suggested that variations in the Earth's orbit around the Sun produced the effect. In the 1920s, Milutin Milankovitch, a Serbian engineer, developed Croll's hypothesis further.

A Theory Takes Shape

The Croll/Milankovitch theory attributes the chilling and warming of the globe to changes in the amount of sunlight hitting the Earth. These changes result from the tilting and wobbling of the Earth on its axis, and the stretching of the path the Earth takes in its journey around the Sun. These cycles repeat every 41,000 years, 23,000 years, and 100,000 years respectively. We could expect climate records to show temperature changes at these intervals.

The history of the Earth's climate is preserved in ice in Greenland and Antarctica, and in sea-floor sediments and coral reefs in the world's oceans. Ice cores can take us back some hundreds of thousands of years, with each year's snowfall laid down on the last, like vertical tree rings. Sea-floor sediment can take us back millions of years.

For most of the last century, the Croll/Milankovitch theory seemed to fit what we knew of the cycle of the ice ages. Recently recovered evidence of ancient climate history, however, doesn't quite match the expectation of the Croll/Milankovitch theory. Of the three effects described by the theory, the wobble of the Earth's axis has the greatest influence on how much sunlight reaches the Earth. The theory predicted ice ages would occur every 23,000 years, mainly as a result of this wobble. In fact, though, the ice and sea-floor records of the last 600,000 years show a clear pattern of 100,000-year recurrences.

While another effect, the stretch of the Earth's orbit from circular to slightly oval, has an approximate 100,000-year period, the resulting difference in sunlight isn't great enough to begin or end an ice age. Supporters of the Croll/Milankovitch theory suggest that the combination of the stretch cycle and some internal process on Earth that increases the effects could create the right conditions at the right times.

Second Theory Clouds the First

Richard Muller, an astrogeophysicist at the University of California at Berkeley, isn't convinced. “There are really two problems,” he says. “One is that the regularity of the 100,000-year cycle is much greater than you would expect from the Croll/Milankovitch theory. And then there's another challenge that we call the causality problem. Recently we've managed to obtain much better dates for the terminations of the ice ages.” Accurate dating of calcite from a cave in Nevada and of sea-floor corals shows that one ice age ended about 135,000 years ago. The Croll/Milankovitch theory says that the warming would have begun about 7,000 years after that. So if this theory were right, the cause — the decrease in sunlight — would have come after the effect — the melting of the ice.

“Back in 1993,” Muller says, “I had a hypothesis that dust from space might be affecting climate. I think it could do this through its effect on cloud cover. Clouds have a big effect on climate.”

He knew that the plane of the Earth's orbit shifts back and forth over time. If you stretch a rope from the center of the Sun to the center of Jupiter, you will find that as the Earth circles the Sun, sometimes it's above that rope and sometimes it's below it. A ring of dust, probably left over from collisions between asteroids and other bodies, also orbits the Sun, on the same plane as Jupiter. When the Earth is above and below that plane, it travels through empty space. But when it orbits in the same plane as the dust, as it does every 100,000 years, we can expect some of that dust to enter the atmosphere.

Muller says, “I predicted that if anybody could ever measure the variations in dust, they would see a 100,000-year cycle that would match the 100,000-year cycle of the climate.” Six months after Muller announced this, a geochemist at Caltech looking at sea-floor sediment cores found increases in extraterrestrial dust at intervals that matched those of interglacials.

The quantity of dust, however, was very small. “We now believe that the amount of dust is something like 15,000 tons per year,” Muller says. “That sounds like a lot. Spread it out over the whole Earth and you have to wait ten years before it's one atom thick. It's hard to see how such a small amount of dust could have such a big effect on the climate.”

He's convinced, however, that the matching 100,000-year cycles of dust and ice are not just coincidence. His space dust theory fits the climate history data very well, and doesn't have the causality problem that the Croll/Milankovitch theory has.

“I think the weakness of my theory is that we don't yet understand what's going on in the atmosphere,” Muller says. “The cloud cover mechanism is actually quite complicated. The dust comes in and interferes with the electrical currents that help create the seeds of clouds.”

Studying the upper atmosphere poses special problems. “It's harder to reach than the oceans,” Muller says. “You can swim down in the ocean or drop things down there. You can reach the lower atmosphere — up to 150,000 feet — by balloons or very high-flying airplanes. With a satellite, you can go up 100 miles or more. But there's that gap in between that's extremely difficult to measure. Many of the measurements are made by rockets that are just passing through it. That's sort of like seeing Paris through the window of a train. You miss a lot.”

Still Looking for the Key

What really causes the ice ages? We still don't know the details. “It's a field in which there's a great deal of opportunity for a young person coming in,” Muller says. “Ancient climate is the key to present climate. To understand it fully, we have to figure out what happened in these huge cycles that changed climate so dramatically.”


In this case, the time interval between two successive occurrences or phases of a recurrent event; a cycle

Back to Top


  1. When did the last ice age end?
    [anno: The last ice age ended about 12,000 years ago.]
  2. When might the next ice age occur?
    [anno: The next ice age could occur anytime from now until about 8,000 years from now.]
  3. Imagine an ice age occurred on the planet. How would life change? How would people survive? How would they grow food and produce energy? What would happen to plants and animals? Write a paragraph about some of the changes the planet might go through if an ice age happened tomorrow. Include details about how people might prepare for this ice age and what the changes to daily life would be like.
    [anno: Answers will vary.]