Much of our understanding of the ancient history of our planet comes from radioisotope dating, a process where scientists calculate the amount of certain isotopes in a geological sample to determine how old it is. But according to new research out of North Carolina State University, a flaw in this widely-used technique may be skewing the results so samples seem much older than they really are.

Radioisotope dating hinges on the fact that over time, certain radioactive isotopes will decay to form other isotopes: For example, rubidium-87 decays into strontium-87, with a half-life of 48.8 billion years, but strontium-87 is stable so it doesn't decay. To estimate when a sample of rock formed, scientists first calculate the concentration of both rubidium-87 and strontium-87 in the sample, and then compare those to the concentration of strontium-86, as a kind of baseline.

By calculating the ratios of rubidium-87 to strontium-86, and strontium-86 to strontium-87, a graph called an isochron is created, which scientists can then use to determine the age of a sample. The process works best on igneous rocks, and has been used to study Earthly and lunar formations for decades. But, as the NC State study suggests, that final figure might not be taking other variables into account.

For one, the atoms of different elements will diffuse through a material at different rates due to a process known as differential mass diffusion. In this case, strontium-86 atoms are smaller than strontium-87 or rubidium, meaning they will spread through surrounding rock faster, and that differential may be influenced further by the properties of the sample itself.

"It's a slow process, but not necessarily a negligible one when you're talking about geological time scales," says Robert Hayes, author of the study. "The rate of diffusion will vary, based on the sample – what type of rock it is, the number of cracks and amount of surface area, and so on."

The process as it's currently applied, Hayes says, is likely to overestimate the age of samples, and considering scientists have been using it for decades, our understanding of Earth's ancient timeline could be worryingly inaccurate. That said, Hayes does point out that the issue doesn't affect carbon dating, a separate process that's used to date younger samples on a scale of thousands of years rather than millions or billions.

"There's not a simple equation that can be applied to every circumstance," says Hayes. "Researchers will need to evaluate samples individually, then apply the relevant physics accordingly. It's a pain in the neck, but it will make our estimates significantly more accurate. If we don't account for differential mass diffusion, we really have no idea how accurate a radioisotope date actually is."

The research is published in the journal Nuclear Technology.