''But at earlier times, the carbon dates were substantially younger than the dates we estimated by uranium-thorium analysis,'' he said.
''The largest deviation, 3,500 years, was obtained for samples that are about 20,000 years old.'' One reason the group believes the uranium-thorium estimates to be more accurate than carbon dating is that they produce better matches between known changes in the Earth's orbit and changes in global glaciation. Fairbanks, a member of the Lamont-Doherty group, said that if the dates of glaciation were determined using the uranium-thorium method, the delay - and the puzzle - disappeared.
In cases where different sources are characterized by the same zircon U-Pb ages, differential metamorphism and/or exhumation may produce distinctive thermochronological ages.
According to carbon dating of fossil animals and plants, the spreading and receding of great ice sheets lagged behind orbital changes by several thousand years, a delay that scientists found hard to explain. The group theorizes that large errors in carbon dating result from fluctuations in the amount of carbon 14 in the air.
Changes in the Earth's magnetic field would change the deflection of cosmic-ray particles streaming toward the Earth from the Sun.
They arrived at this conclusion by comparing age estimates obtained using two different methods - analysis of radioactive carbon in a sample and determination of the ratio of uranium to thorium in the sample.
In some cases, the latter ratio appears to be a much more accurate gauge of age than the customary method of carbon dating, the scientists said.Carbon 14 is thought to be mainly a product of bombardment of the atmosphere by cosmic rays, so cosmic ray intensity would affect the amount of carbon 14 in the environment at any given time.