Antarctic Rocks Reveal Hidden Flaws in the Clocks That Track Ice Sheet Retreat

Deep in West Antarctica, beside the glacier that scientists fear most, a volcanic mountain has been quietly keeping a diary of the ice that once buried it. Mount Murphy rises above Thwaites Glacier in the Amundsen Sea Embayment, and the boulders scattered across its flanks carry an atomic record of when the ice last retreated. A new study of these rocks, published in the journal Geochronology, has turned that record into something far more valuable than a simple history lesson: a stress test of the dating methods themselves, one that exposes subtle weaknesses in the isotopic clocks used to reconstruct how fast ice sheets collapse.

The research, led by Jonathan Adams of Imperial College London and the British Antarctic Survey with colleagues including Dylan Rood, Klaus Wilcken, Stephen Roberts, and Joanne Johnson, hinges on a technique called cosmogenic nuclide surface exposure dating. The principle is elegantly simple. When cosmic rays strike rock at the Earth’s surface, they shatter atoms in the mineral quartz and forge rare isotopes such as beryllium-10 and carbon-14. The longer a rock sits exposed to the sky, the more of these cosmogenic nuclides accumulate. Measure their concentrations, and you can calculate how long ago the ice melted away and the rock first saw daylight. For ice sheet scientists, these ages are the ground truth against which computer models of past and future retreat are validated.

But the two isotopes behave very differently, and that difference is the heart of the study. Beryllium-10 has a half-life of 1.387 million years, so it accumulates essentially forever and lingers through multiple cycles of burial and re-exposure. In Antarctica, where cold-based ice often freezes to the bed without eroding it, rocks can carry inherited beryllium-10 from exposures that happened tens or even hundreds of thousands of years ago, long before the last glacial maximum. That inherited signal can make a freshly deglaciated surface look deceptively old. In situ produced carbon-14, by contrast, has a half-life of just 5,700 years. Any carbon-14 built up before the last glacial maximum largely decays away while the rock sits under deep ice, so a surface re-exposed at the end of the ice age carries almost none of this legacy. A rock that was saturated with carbon-14 before the last glacial maximum, buried at 25,000 years ago, and re-exposed at 10,000 years ago would retain a pre-ice-age inventory amounting to only about six percent of what is measured today. That makes carbon-14 uniquely suited to dating the most recent chapter of ice retreat.

Adams and his team collected nine rounded erratic cobbles from three sites on Mount Murphy: the Notebook Cliffs at around 850 to 900 meters above sea level, Turtle Rock between roughly 440 and 700 meters, and a scoria cone near Kay Peak sitting just 180 to 240 meters above the ice, less than a kilometer from the grounding line of Pope Glacier. All nine had previously been dated with beryllium-10, and those ages told a coherent story: the ice surface at Mount Murphy thinned dramatically during the early to mid-Holocene, reaching low elevations by roughly 9,000 to 6,000 years ago. The new carbon-14 measurements, however, told a different tale. Six of the nine samples yielded carbon-14 ages of only 5,000 to 3,000 years, implying the ice lingered thousands of years longer than the beryllium-10 record suggested.

When the two isotopes from the same sample are compared on a paired nuclide diagram, where the beryllium-10 concentration is plotted against the carbon-14 to beryllium-10 ratio, the samples split into two camps. Concordant pairs, in which the two ages agree within uncertainty, indicate a simple exposure history since the ice age. Discordant pairs, where the ages do not overlap, hint at burial, shielding, or shifting production rates. At Mount Murphy, some samples were concordant while others, often at the same elevation, were discordant, and the discordant carbon-14 ages were systematically the younger ones. Stranger still, some high-elevation samples yielded younger ages than low-elevation ones, the exact inverse of the pattern expected as an ice sheet thins upward through time.

The team systematically hunted for a geological explanation. They examined aerial photographs from 1966 showing that the lower scoria cone outcrop was once buried by ice, confirming that at least some samples experienced late Holocene shielding. They tested whether thin layers of till or snow could explain the discrepancies, and whether erratics might have been exposed at higher elevations before being transported downslope, a mechanism invoked at Shark Fin Nunatak near Tucker Glacier. They even modeled scenarios in which subsurface production of carbon-14 by muons, which contributes roughly twenty percent of total carbon-14 production at the surface compared with under two percent for beryllium-10, could inflate the carbon-14 to beryllium-10 ratio in buried samples, producing seemingly impossible nuclide ratios like those seen at Sjögren Glacier on the Antarctic Peninsula. None of these geological scenarios could account for the Mount Murphy pattern.

The breakthrough came from the laboratory rather than the landscape. The researchers had run repeat carbon-14 measurements on four of their samples, and three of those four failed to reproduce within their stated uncertainties. The initial measurements produced ages 5,000 to 3,000 years younger than the replicates, with some replicate ages more than 150 percent older than their first-pass counterparts. Crucially, the older, reproducible ages agreed with the beryllium-10 chronology, while the young, irreproducible ages did not. Sensitivity analyses using different blank corrections and different calibration values for the CRONUS-A reference material could not rescue the young ages, no matter how the numbers were shuffled. The problem, it appeared, lay in the measurements themselves, not in the ice history.

To find out whether this was an isolated glitch, the team turned to the informal cosmogenic-nuclide exposure-age database, ICE-D, and compiled every sample worldwide with replicate carbon-14 measurements, thirty-one in total, spanning the Antarctic Peninsula, the Weddell Sea Embayment, Lake Bonneville in Utah, and the Northwest Highlands of Scotland. The result was sobering: in eighteen of the thirty-one samples, at least one carbon-14 measurement failed to reproduce within the nominal six percent uncertainty conventionally assigned to such data, and fifteen still failed at the more forgiving two-sigma level. Interlaboratory comparisons of CRONUS-A tell a similar story, with long-term average carbon-14 concentrations varying by roughly fifteen percent between extraction facilities, from about 6.12 times ten to the fifth atoms per gram at Tulane University to 7.28 times ten to the fifth at ETH Zurich. The inescapable conclusion is that the analytical uncertainty quoted for in situ carbon-14 is currently underestimated, particularly for the moderate concentrations typical of Holocene-aged samples, where contamination from background carbon in processing blanks becomes dominant.

The good news for Antarctic science is that the accepted deglaciation history of Mount Murphy, and by extension the Amundsen Sea Embayment, survives this scrutiny intact. The concordant, reproducible carbon-14 ages align with the beryllium-10 record, confirming that ice there thinned substantially between about 9,000 and 6,000 years ago, and the team credits access to the Tulane laboratory’s quality control data with preventing them from publishing a badly wrong chronology. The broader lesson is a call to arms for the geochronology community: routinely publish replicate measurements, quantify and minimize scatter in process blanks, and pursue interlaboratory comparison studies with additional reference materials to establish consensus calibration values. As automation of carbon-14 extraction lines increases throughput, these checks become practical. The stakes could hardly be higher. Thwaites Glacier is one of the largest contributors to sea-level rise from Antarctica, and projections of its future depend on models calibrated against the past. If the isotopic clocks that record that past are even slightly miscalibrated, the sub-millennial precision needed to capture phases of retreat and readvance slips out of reach. This study shows both the fragility and the resilience of the method: the clocks can go wrong, but with enough cross-checks, science can catch them before the story gets rewritten.

Subject of Research: Cosmogenic carbon-14 and beryllium-10 exposure dating of Holocene ice sheet thinning at Mount Murphy, West Antarctica

Article Title: Paired 14C–10Be exposure ages from Mount Murphy, West Antarctica: Implications for accurate and precise deglacial chronologies

Article References: Adams, J. R., Rood, D. H., Wilcken, K., Roberts, S. J., & Johnson, J. S. (2026). Paired 14 C– 10 Be exposure ages from Mount Murphy, West Antarctica: Implications for accurate and precise deglacial chronologies. Geochronology, 8(2), 255-277. https://doi.org/10.5194/gchron-8-255-2026

Image Credits: AI Generated

DOI: 10.5194/gchron-8-255-2026

Keywords: cosmogenic nuclides, in situ carbon-14, beryllium-10, Mount Murphy, Thwaites Glacier, West Antarctica, exposure dating, deglaciation, Holocene, ice sheet thinning, sea-level rise, analytical uncertainty