The imprints of raindrops preserved in 2.7bn-year-old rock are being used to figure out what the atmosphere was like on the early Earth.
Scientists have used the depressions drops left to calculate how fast they were going as they impacted the ground.
This has allowed them to determine the density of air in ancient times.
This palaeobarometric approach, revealed at the AGU Fall Meeting, will help constrain the models that try to simulate conditions in Archaean times.
Earth 2.7 billion years ago was very different from the planet we know today.
It turned much faster, the Moon was closer and the Sun was much weaker. And there were no animals or plants in existence back then; the air was simply not breathable.
"There was probably quite a bit of nitrogen in the atmosphere, like today, but there was no oxygen," explained Sanjoy Som from Nasa's Ames Research Center.
"The oxygen was likely replaced by greenhouse gases such as carbon dioxide and methane.
"My palaeobarometry work cannot tell you precisely what the gases were, but it will assist modellers of atmospheric composition by giving them a constraint," he told BBC News.
Dr Som told the AGU meeting - the largest annual gathering of Earth scientists - that the "fossil raindrops" were discovered in Ventersdorp in the North West Province of South Africa in the 1980s.
They consist of lots of pits in the surface of a rock that started out as volcanic ash-fall.
Rain tumbling on to the ash would have dug out small depressions, which were then covered over by further ash deposits and lithified, or turned to stone.
We only see the imprints today because the top layers of the rock have now been eroded back.
Dr Som's and colleagues' thinking is that the pits should tell us something about ancient air pressure.
Their starting point is that the diameters of the imprints are controlled ultimately by the top speed of the raindrops as they hit the ground.
This number - the terminal velocity - is dependent on air density. In the modern atmosphere it is about 9m/s.
"The rationale here is that if the air back then was thicker, the raindrops would fall slower, and the craters in the ash would be smaller; and conversely, if the air was thinner, the drops would fall faster and the craters would be larger," said Dr Som, who is also affiliated to the Blue Marble Space Institute of Science, Seattle.
The confounding factor would be if raindrops were somehow much bigger in Archaean times. Fortunately, it turns out the maximum size a raindrop can reach is independent of air density; it is controlled by aerodynamic forces that are unrelated to the thickness of the air.
The fattest drops 2.7bn years ago would have been the same as they are today - about 7mm.
Dr Som's team conducted experiments in which, using a pipette, they dripped small, carefully controlled volumes of water into a tray of volcanic ash from a height of about 25m.
This allowed the group to relate the momentum of a raindrop to the size of the imprints made; and then, using theory, to calculate the momentum of a drop of a known size at any air density.
The team concluded that if the biggest imprints in the Ventersdorp rock were formed by the largest raindrops, air pressure in the Archaean could have been no more than twice what it is today.
"But knowing what we do about the spread of raindrop sizes, we know the largest possible size is actually quite rare.
"So if it was smaller raindrops that formed the largest imprints at Ventersdorp then the atmospheric density was probably similar to ours, if not less."
The study supports the idea therefore that the ancient atmosphere must have had a strong concentration of greenhouse gases.
If air pressure were the same or even lower than it is today, there is no other way to explain why Earth was not thrown into snowball conditions by a substantially weaker Sun.
Without extra thickness in the atmosphere to trap heat, the properties of the gases themselves had to provide the blanket.
Jonathan.Amos-INTERNET@bbc.co.uk and follow me on Twitter: @BBCAmos