The part of this story that stayed with me was not the depth, although 2.9 kilometres of rock is difficult enough to picture. It was the taste.
Barbara Sherwood Lollar, a geochemist at the University of Toronto, has spent decades studying water that flows from fractures in the Kidd Creek mine near Timmins, Ontario. Drops of the ancient brine landed on her as it bubbled from the rock. That was enough to register its intense saltiness and bitterness.
She did not fill a glass and drink it. That is one of several details that have become blurred as the story has travelled online.
I have a one-year-old daughter, so meals in our home often become small experiments in taste. Reading about Sherwood Lollar, I kept thinking about that ordinary sense applied to something so far outside ordinary time: a human tongue briefly registering chemistry that had developed in darkness for longer than animals or land plants have existed.
The image is powerful, but the science is more precise than the viral version. The water was not found in an underground lake, and nobody directly measured the birthday of each molecule. Researchers analysed gases dissolved in brine moving through tiny fractures in ancient rock. Their result was an estimated mean residence time, with a substantial range and some necessary assumptions.
The deeper samples came from 2.9 kilometres down
Kidd Creek is a copper-zinc mine cut into the Canadian Shield, the vast region of very old continental rock surrounding Hudson Bay. The ore deposit and its host rocks formed roughly 2.7 billion years ago, when the geography and atmosphere of Earth looked nothing like they do now.
In 2013, a team led by Greg Holland reported on fracture fluids collected 2.4 kilometres below the surface. Their paper in Nature concluded that some fluid components had been isolated for at least about 1.5 billion years, with different dating models producing older estimates.
Mining later opened access to a deeper level. For a 2018 study in Geochimica et Cosmochimica Acta, Oliver Warr and colleagues analysed fracture fluids from 2.9 kilometres down. Estimates from different noble gases placed their mean residence times between about 1.0 and 2.2 billion years. The authors considered an average near 1.7 billion years the best overall description.
That is where “roughly two billion years” comes from. It is a reasonable summary, but not an exact date.
The clock was hidden in noble gases
There is no carbon dating here. Radiocarbon dating works over tens of thousands of years, not billions.
The team instead measured helium, neon, argon, krypton and xenon. These noble gases are chemically reluctant to react, which makes their isotopes useful tracers. Some accumulate as radioactive elements in the surrounding rock decay. Others can preserve information about ancient atmospheric sources.
The longer a fluid system remains cut off from the surface, the more radiogenic gas can build up. By comparing the measured amounts and isotope ratios with production models for the rock, researchers can estimate how long the fluid has resided in the crust.
Different gases do not produce one perfectly matching answer. They are generated at different rates, can move differently, and depend on assumptions about the history and composition of the rock. That is why the 2018 paper reports a range rather than a single birthday.
“Mean residence time” also allows for mixing. A sample can contain components with different histories, just as the average age of people in a room does not tell you the age of every person. What the measurements strongly indicate is that part of this deep fracture system has remained extraordinarily isolated on a geological timescale.
The famous taste was a splash, not a drink
The taste story needs two careful distinctions. The first is depth. Sherwood Lollar discussed tasting the brine in a 2013 interview with the Los Angeles Times, after publication of the 2.4-kilometre study. The nearly three-kilometre samples were analysed later.
It is therefore safer to say that she encountered the taste of ancient Kidd Creek fracture water, not that she definitely tasted the precise 2.9-kilometre sample dated in the 2018 paper. It is the same mine and the same broad class of deep brine, but the researchers found that the two depths belonged to hydrogeologically distinct systems.
The second distinction is between tasting and drinking. The old interview helped create the impression that Sherwood Lollar deliberately sipped the water. In a 2026 clarification to IFLScience, she said she did not drink it. The water flows and bubbles from the fractures, so droplets inevitably land on researchers. Those splashes were enough for her to recognise that it was bitter and many times saltier than seawater.
The sensory description tells us something useful about the chemistry. The brine is slightly viscous and clear when it emerges. Contact with oxygen causes dissolved minerals, especially iron, to form an orange colour.
A 2016 analysis in Nature Communications reported that deep Canadian Shield fracture waters can contain as much as 325 grams of dissolved solids per litre. They are rich in calcium, sodium, magnesium and chloride. This is not ancient drinking water waiting to be bottled. It is chemically aggressive brine produced by a long history of water-rock interaction.
The last surface world it knew was mostly microbial
Two billion years ago does not mean Earth was lifeless. Microbes had already existed for well over a billion years. Photosynthetic organisms had begun transforming the atmosphere, and the long rise of oxygen was under way.
But the familiar living world had not arrived. The Smithsonian’s history of early life places the evolution of the first animals more than a billion years later. Land plants, forests, insects, dinosaurs, mammals and humans were farther away still.
There is another subtlety here. The age estimate does not prove that this particular water once sat at the surface two billion years ago and then slipped into a crack. Some of the fluid may descend from ancient seawater or later groundwater, while some water molecules and dissolved compounds may have been created or altered through reactions in the rock.
The result dates the isolation and evolution of a fluid system. It gives us a chemical archive of the deep crust, not a sealed flask of an ancient ocean.
Isolation did not necessarily mean sterility
Sunlight cannot reach this environment, but energy can be made underground. Natural radioactive decay in the rock can split water molecules through radiolysis, producing hydrogen and reactive compounds. Water reacting with minerals can generate more hydrogen, methane and simple organic molecules.
The 2016 sulphur study found a pathway that could also produce sulphate from sulphide minerals. Together, hydrogen and sulphate provide the two sides of a chemical reaction that certain microorganisms can use for energy without sunlight.
More recent work has moved beyond asking whether Kidd Creek is merely habitable. A 2024 study in Communications Earth & Environment used genetic, lipid and isotope evidence to identify an indigenous microbial community in fracture water from 2.39 kilometres down. Samples isolated from the mine environment were dominated by a salt-tolerant, oxygen-avoiding bacterium the researchers called Candidatus Frackibacter.
That finding does not show that the organisms themselves are two billion years old. The 2024 team explicitly noted that the microbes could be as young as the youngest water mixed into the fracture system. Contamination is also a serious problem in any working mine, which is why the researchers compared their deep samples with mine water and microbes from exposed surfaces.
What they found is narrower and, to me, more interesting: the deep crust is not just inert rock with old water trapped inside. It can be a slow chemical system, making energy and supporting sparse life on timescales almost detached from the surface.
The taste anecdote gives the story its human scale. The noble gases give it its scientific weight. For a moment, those two scales met in one encounter: the timescale of a tongue, measured in seconds, and the timescale of a hidden water cycle, measured in billions of years.
