The ice age is coming" sang the Clash in the 70s, and headlines shouted, "Brace yourself for another ice age."

Even lyrics of pop songs come from somewhere, and these arose because some thought that the world was facing a frozen future. In an era of global warming, the idea of ice sheets covering half of Europe in the next few centuries seems ridiculous. But some scientists in the early days of climate modelling were anticipating a colder world.

The story begins in the early 20th century with a Serbian astrophysicist, Milutin Milankovitch, who believed that wobbles in the earth's orbit around the sun had a profound effect on the earth's temperature and devoted his career to developing a theory based on these principles. Milankovitch predicted that ice ages occurred at particular orbital variations of the earth, when less solar energy reached its surface at northerly latitudes.

From the position of the earth's orbit he calculated the temperature 600,000 years ago. After initial enthusiasm, no one paid much attention to Milankovitch's ideas until the late 60s, when a study of deep sea sediments showed that ice ages occurred at times which fitted in with Milankovitch's theory.

Patterns in the earth's orbit can be predicted accurately. So could the next ice age be foretold? The exact orbital conditions which give rise to cool summers in northern latitudes (thought to trigger ice ages) are uncertain. As William Connolley, climate modeller with the British Antarctic Survey says: "The connection between solar radiation at certain latitudes and the onset of ice ages is not as certain as people once thought."

The earth has probably already passed the peak of the current interglacial period, and if Milankovitch's rules are followed - leaving aside human influence - then a trend towards a cooler climate would be expected within the next 1,000 years. But even 30 years ago, the powerful warming effect of atmospheric carbon dioxide was well known. But CO2 has rivals in the greenhouse world.

Aerosols thrust into the atmosphere from the very same process that generates CO2 - the burning of fossil fuels - act as atmospheric coolants, reflecting solar energy back into space. At the time it was not known which would control atmospheric temperatures.

The "cooling" argument was backed by temperature measurements from the northern hemisphere which showed a cooling trend from the 40s to the 70s. Armed with 100 years of temperature readings from around the world, climate scientists of the 70s tried to estimate how hot the world would be in future. Whether they predicted a rise or fall in temperature depended on the emphasis given to CO <->2 or aerosols, and on the timescale of the calculation - Milankovitch cycles cover huge time periods, so predictions based on orbital cycles with an accuracy closer than 2000 years are not possible.

Some stark claims were made. In 1971 the journal Science reported that if atmospheric aerosol concentrations increased eightfold, then the subsequent cooling "if sustained over a period of several years _ is believed to be sufficient to trigger an ice age." The threat of a nuclear winter at that time - with dust from an atomic explosion blocking out the sun - weighed heavily on the public's mind, so it is easy to see where headlines about a future ice age came from. Dr Connolley has studied the issue of whether an imminent ice age was predicted in the 70s and believes that the argument depends on how you define "imminent".

"If you work on greenhouse-type timescales then 'imminent' means the next century or so," he says. "In this case, the scientific papers from the 70s show no evidence of predictions of an 'imminent' ice age." Nigel Calder, former editor of New Scientist and atmospheric researcher disagrees. In his book The Weather Machine, published in 1974, he wrote: "One might argue that there is a virtual certainty of the next ice age starting some time in the next 2000 years. Then the odds are only about 20-to-1 against it beginning in the next 100 years."

As the debate continued, it became clear that the greenhouse effect of CO <->2 overrides the cooling force of aerosols. The 30-year northern hemisphere cooling trend stopped in the 70s, and scientists realised that if temperatures of the southern hemisphere were included in their models, then no global pattern of cooling was observed anyway. Does the fact that some scientists predicted the wrong trend in the 70s mean that the international consensus about the current warming trend could be misguided too?

"The volume of evidence for the greenhouse effect today far exceeds that available for global cooling in the 70s," says Dr Connolley. "You didn't find government ministers going to a meeting about global cooling like they went to the Kyoto global warming summit." But the "Ice Age is Coming" headlines of the 70s have an important modern resonance - sceptics use them to discredit global warming theories. In the 1999 Reith Lecture, Anthony Giddens said that "only about 25 or so years ago, orthodox scientific opinion was that the world was in a phase of cooling. Much the same evidence that was deployed to support the hypothesis of global cooling is brought into play to bolster that of global warming - heat waves, cold spells, unusual types of weather."

While Prof Giddens has a point, it is untrue to suggest that the scientific consensus about global cooling was anything like that in favour of global warming today. And the next ice age? The earth has been oscillating between long glacial periods about every 100,000 years. Will it do so again? Dr Connolley says, "Without any human interference, there would certainly be an ice age at some point in the future. But inputs of carbon dioxide to the atmosphere have almost certainly overridden any effects of orbit-related cooling for the next few hundred years."

• Dr Alison George is with the British Antarctic Survey.