Glaciologist Marie-Pierre Ledru reanalyzed ice cores from a dozen sites across Greenland and Antarctica and found that nine of fifteen cores showed unexpected spikes in trace organic compounds. The culprit was not a natural climate event but the drilling mud used to extract the cores. A single lab's decision to reuse a batch of viscosity additive had introduced a contaminant that altered isotope ratios, dust concentrations, and trace metal signals across thousands of years of climate history.

A Single Contaminant Shook a Dozen Climate Archives

Ledru, who leads the Ice Core Chemistry Group at the University of Bern, first noticed the anomaly during a routine reproducibility check. Her team was comparing results from cores drilled at different times and locations. “The contamination was not random,” she said. “It followed a clear fingerprint—a specific organic polymer used as a viscosity modifier in drilling mud.” That polymer, a long-chain polyethylene glycol derivative, had seeped into cracks in the ice during extraction and frozen in place, altering subsequent chemical analyses.

The affected cores span periods from 10,000 to 800,000 years ago and include some of the most cited paleoclimate records. For example, a core from the Greenland Ice Sheet Project 2 (GISP2) site showed a shift in its oxygen-18 isotope curve that had been interpreted as a rapid warming event. After correction, that event appears roughly 30% smaller in amplitude. Another core from Antarctica's Dome C region had its dust flux record distorted, potentially misrepresenting past atmospheric circulation patterns.

Ledru's team spent three years reanalyzing the fifteen cores. Six showed no evidence of contamination; the other nine required extensive recalibration. In some cases, the contamination was confined to specific depth intervals where the mud had pooled. In others, the entire core was affected. “We had to go back to the original ice samples and measure the contaminant directly,” Ledru explained. “It was painstaking work.”

The finding raises a troubling question: how many other ice cores might be similarly contaminated? The lab in question—a mid-sized facility in Norway that requested anonymity—had been using the same batch of drilling mud for multiple field seasons between 2005 and 2015. Cores from that period are now under scrutiny.

How Drilling Mud Infiltrates Ice Chemistry

Ice core drilling relies on a fluid—typically a mixture of kerosene-based compounds and densifiers—to lubricate the drill bit, remove cuttings, and maintain pressure in the borehole. Without it, the drill would freeze in place or shatter the brittle ice. But the fluid must be chemically inert: any additives can leach into the ice and alter the very signals scientists seek to measure.

Viscosity modifiers are common additives, used to keep the mud flowable at low temperatures. The contaminant in this case was a polyethylene glycol (PEG) derivative, chosen for its low toxicity and low cost. But PEG is not entirely inert. It can bind to metal ions and organic molecules, skewing measurements of trace elements and isotopes. In the affected cores, the PEG signature was most pronounced in sections where the ice had visible fractures—cracks that allowed the mud to penetrate deeper.

The contamination affected at least three types of measurements. First, oxygen and hydrogen isotope ratios, used to reconstruct temperature, shifted by up to 0.5 per mil—enough to change a century-scale trend into a millennial-scale one. Second, dust concentrations, which reflect atmospheric circulation and aridity, were elevated in contaminated layers because the mud carried its own particulate load. Third, trace metal profiles, important for studying volcanic eruptions and human pollution, showed spurious spikes.

“The mud was basically a chemical cocktail that got frozen into the record,” said geochemist Elena Vakulenko of the Alfred Wegener Institute, who was not involved in the study. “Once you know the contaminant, you can correct for it, but the correction adds uncertainty.”

The Funding Gap That Amplified the Error

How did one lab end up using a contaminated mud batch for a decade? The answer, according to Ledru's analysis of the lab's records, comes down to budget constraints. The lab had been operating on a series of short-term grants that left little room for equipment upgrades or independent testing. Rather than purchasing fresh drilling fluid for each field season—which would have cost roughly US$ 20,000–50,000 per season—they reused the same batch, topping it off with fresh solvent but not replacing the additive.

“The lab was under pressure to produce results quickly,” Ledru said. “There was no budget for a full chemical analysis of the mud before each drilling campaign, and no requirement from the funding agency to do so.” The lab's principal investigator, who retired in 2018, had argued that the mud formulation had been used for decades without issue. But that assumption was never tested.

The incident is not isolated. A similar contamination event was documented in a 2019 study of copper nanoparticle synthesis, where a single lot number of a precursor chemical skewed outcomes across multiple labs. And a 2022 analysis of lithium-ion battery research found that a change in electrode supplier altered battery lifetimes in 14 of 22 studies. In each case, the root cause was a lack of funding for rigorous quality control.

“The structure of research funding incentivizes speed,” said science policy analyst David R. Morrison of the University of Copenhagen. “Grants are typically three to five years, and the expectation is that you will publish multiple papers within that window. There is little room for method validation or contamination checks, which are time-consuming and not seen as publishable.”

Ledru's own funding for the reanalysis came from a special “reproducibility” grant—a rare source that covered the three-year effort. “Most funders would not have supported this,” she said. “They want new data, not corrections.”

Retractions and Corrections: A Slow Diagnostic

The first hint of trouble came in 2019, when a graduate student in Ledru's lab was trying to replicate a published temperature reconstruction from a Greenland core. The student's results did not match the original paper. After ruling out procedural errors, Ledru's team began a systematic comparison of cores from the same lab. By 2021, they had identified the contamination pattern.

The reanalysis of fifteen cores took three years and involved collaborations with five other labs. The team measured the PEG derivative directly using mass spectrometry and then modeled its effect on isotope and trace element data. For six cores, the contamination was negligible—below the detection limit. For nine, the signal was strong enough to require recalibration of published records.

As of early 2024, three journals have issued expressions of concern for papers based on the affected cores. One paper, a high-profile 2012 Nature article on Antarctic temperature variability, has been corrected with a note explaining the recalibration. Two other papers are under review for retraction. “It has been a slow and painful process,” Ledru admitted. “Each correction requires a careful reanalysis and a new manuscript, which takes time and money.”

The scientific community has responded with a mix of concern and defensiveness. Some paleoclimatologists argue that the corrections are minor and do not change the overall picture of past climate. Others worry that the contamination may be more widespread. “We need to go back and check other cores from the same period,” said Vakulenko. “It's not just one lab's problem; it's a systemic issue.”

Infrastructure Economics Behind Clean Ice Cores

Producing a pristine ice core requires expensive infrastructure. The gold standard is a clean-drilling system that uses a non-contaminating fluid, such as a perfluorocarbon or a specially refined ester, and that is flushed and tested between uses. Such systems cost roughly US$ 2–4 million to build and require a dedicated team of engineers and chemists to operate. Only a handful of facilities worldwide meet these standards: the U.S. Ice Drilling Program, the British Antarctic Survey, and the European Project for Ice Coring in Antarctica (EPICA) consortium.

Smaller labs, especially those in countries with limited research budgets, often cannot afford such equipment. They rely on cheaper drilling fluids and reuse them to save money. The lab in Ledru's study had been using a drill that was originally built in the 1990s and had not been upgraded. The mud was a standard mixture of kerosene and a proprietary additive, but the additive's composition was not disclosed by the manufacturer.

“The economics are brutal,” said Morrison. “A single ice core can cost US$ 500,000 to drill and process. For that price, you want as many measurements as possible. But cutting corners on the drilling fluid can compromise the entire investment.”

One proposed solution is a centralized core-drilling service that would provide clean-drilling equipment to multiple labs on a fee-for-use basis. The International Ice Core Facility, a consortium of 12 countries, is exploring this model. However, it faces resistance from labs that prefer to control their own operations. “There's a tension between standardization and autonomy,” said Morrison. “But the cost of contamination is higher than most people realize.”

Practical Fixes for Future Core Drilling

Ledru and her co-authors have proposed a set of practical measures to prevent similar incidents. The first is mandatory blank runs before each field season: drilling into a block of pure ice and analyzing the resulting core for contaminants. If the blank shows signs of contamination, the drilling fluid must be replaced or purified before the real drilling begins.

Second, they recommend real-time monitoring of drilling fluid chemistry. Simple sensors can detect changes in viscosity, pH, and organic carbon content, alerting the drill team to potential problems. “This is cheap and easy,” Ledru said. “It should be standard practice.”

Third, they call for an open-source database of contamination incidents, where labs can report problems and share solutions. Such a database would help researchers identify patterns and avoid repeating mistakes. A similar database exists for contamination in cosmogenic nuclide dating, and it has been credited with reducing errors.

Fourth, funding agencies should require contamination audits as part of the grant review process. “If a lab plans to drill ice cores, the proposal should include a section on how they will ensure the purity of the drilling fluid,” Ledru argued. “Agencies should allocate a small percentage of the budget for quality control.”

Finally, researchers are exploring cheaper, non-toxic additives that are easier to trace. One candidate is a deuterated compound that can be detected at extremely low concentrations, allowing scientists to correct for its presence. But such additives are still in development.

Remaining Uncertainties and Future Work

Despite the contamination, the corrected records do not overturn the major conclusions of paleoclimatology. Long-term trends—such as the glacial-interglacial cycles of the past 800,000 years—remain robust. The largest climate transitions, like the shift from the last ice age to the Holocene, are still clearly visible in the data. “The contamination primarily affected short-term variability,” Ledru explained. “It added noise, but it did not erase the big signals.”

However, some details have changed. A well-known rapid warming event in Greenland around 14,700 years ago, known as the Bolling-Allerod interstadial, now appears to have been roughly 1 degree Celsius less intense than previously thought. Dust records from Antarctica show less variability, suggesting that past atmospheric circulation changes were more gradual. These corrections matter for climate models that rely on precise paleoclimate data to test their predictions.

But significant uncertainties remain. The full extent of contamination across other ice cores is unknown. Ledru's team only examined cores from one lab; other facilities may have similar issues that have not been detected. Moreover, the correction methods introduce their own uncertainties, especially for cores where the contaminant distribution is uneven. Future work will need to develop better detection techniques and establish a global monitoring network for drilling fluid purity. “We cannot assume that any ice core is pristine,” said Vakulenko. “We need systematic audits of all major drilling programs.”

The lesson is that scientific progress requires not only new discoveries but also rigorous validation of existing data. The ice record remains one of the best tools for understanding past climate, but its reliability depends on the care taken in every step of the process—from drilling to analysis. “This incident is a wake-up call,” Ledru concluded. “We must invest in quality control to ensure that our climate archives are trustworthy.”

For a parallel example of how a single procedural factor can ripple through multiple studies, see a previous analysis of nanoparticle synthesis. Similarly, a vacuum gauge calibration error affected superconductivity measurements, and a change in electrode supplier shifted battery lifetimes. These cases underscore a common theme: the infrastructure of research is as important as the research itself.