In August 1987, a scandal erupted over work on the world’s highest mountain range. John Talent, a palaeontologist at Macquarie University in Sydney, Australia, announced he had found numerous anomalies in research charting the geology of the Himalayas. All of the problems were in papers by Vishwa Jit Gupta, one of India’s most celebrated palaeontologists at the time.

More than 10,000 research papers were retracted in 2023 — a new record

Gupta was outraged. But further investigations over the following seven years revealed the scale and brazenness of his fraud. Littered through more than 450 publications were maps of collection sites that didn’t exist, plagiarized images and specimens stolen from collections that Gupta had visited. More than 100 experts had been duped into co-authoring descriptions of fossils Gupta claimed he had collected from specific Himalayan geological formations, but instead hailed from North America and Morocco.

“It destroyed the image of Indian science,” says retired geologist Ghulam Bhat, who completed his PhD on Himalayan geology in the same year that the allegations of fraud emerged.

Gupta died in 2022, and many geologists know to steer clear of his work. But awareness of the problem is fading, and none of Gupta’s papers have been retracted. When Nature contacted Bhat, who is now editor-in-chief of the Journal of the Geological Society of India (JGSI), he was unaware that Gupta’s papers were not marked as retracted. That’s despite the JGSI’s 1991 evaluation of 19 of Gupta’s papers that it published¹, which concluded that “the fossil finds of V. J. Gupta are not reliable [...] the data is incomplete bordering on disinformation”.

Much of Gupta’s work is inaccessible, having been published in now-defunct periodicals or in volumes that, although digitized online, have not been indexed in reference bibliometric databases such as Scopus, Web of Science and Dimensions. Despite this, Gupta’s falsehoods are now being propagated outside specialist circles, as his old journal articles are being surfaced through artificial-intelligence tools that summarize search results. When someone searches for geological landmarks in the Himalayas, such as the Tandi Limestone, Google’s AI overview obliges with direct links to Gupta’s discredited papers.

Bhat now plans to rectify the digital record and says that all other journals that have published Gupta’s work should do the same. “If this problem is left as it is, unattended, it will pollute all of the geological literature in future,” he says. Gupta also published papers in Nature in the 1960s; Chris Graf, director of research integrity at Springer Nature, didn’t confirm that the journal would re-examine them. But he added that in general, when investigations of old papers are hampered by limited available information, an editorial expression of concern “is likely to be our preferred mechanism for alerting readers to potential issues”. (Nature’s news team is editorially independent of its publisher, Springer Nature.)

The Himalayas case raises a wider issue: how to deal with fraudulent, unethical or even merely incorrect research from decades ago, especially in an era in which much old work is digitized and AI systems can pick up studies that might previously have been neglected.

There are many skeletons in science’s closet, but publishers can be reluctant to judge science of the past by today’s standards, and can find it more difficult to issue retractions when authors are dead, original records cannot be consulted or journals are no longer in print.

And as publishers struggle to deal with the fallout from rising research fraud and AI slop, finite resources can also mean that many old papers with known flaws are left as they are.

Gráinne McNamara, a council member of the Committee on Publication Ethics (COPE) in Eastleigh, UK, which advises journals on best practice, says that journals are increasingly contacting COPE to ask for advice about dealing with concerns on very old papers.

“Everybody is looking for what’s the best, most responsible way to handle these,” she says.

Mind the gap

The widespread use of retractions as a mechanism for maintaining integrity in the scientific record is a relatively recent phenomenon. The JGSI evaluation report about Gupta served as a warning notice to the community at the time, but didn’t use the word retraction. It also pre-dates the modern standardization of retraction processes, which happened in the 2000s. Now, a journal should issue a retraction notice, and leave the original paper intact, but with a modified title and a stamp or watermark indicating that it has been retracted. The information should then feed through to bibliographic databases.

In the early 2000s, only around 30 retraction notices were published each year, and COPE first developed guidelines to standardize the retraction process in 2004. The number — and rate — of retractions has exploded since then, with more than 10,000 recorded in 2023 alone.

Digitization of old records, combined with a wider propensity to retract flawed work, has seen journals reaching farther back in time to retract studies — and this has led to some startlingly long gaps between when an article was published and when it was formally retracted. Retraction Watch, an initiative that curates a database of more than 70,000 retractions and expressions of concern (a notice that highlights potential issues with a publication), has reported that the longest gap is 80 years. That’s for a prank case report published in 1923 in the Dutch medical journal Nederlands Tijdschrift voor Geneeskunde.

It described a man who died from “piss cough”, an affliction that caused him to cough up liquid smelling of urine. In 2003, an editor at the journal happened to look up the case, which he thought was widely known to be a hoax. Realizing it had never been retracted and that others did not know it was fake, he sought confirmation of the hoax, and found it in the autobiography of a retired University of Amsterdam physician, who had helped to cook up the nonsense case when he was a medical student. Accompanying commentary on the journal’s website said that a retraction was appropriate now that the older literature had become accessible through digitization.

According to Nature’s analysis of Retraction Watch data — which capture a subset of total retractions — the number of old papers that are being retracted is rising. Overall, some 358 retraction notices and 120 expressions of concern have been issued for papers or book chapters that were more than 20 years old at the time of the editorial action (see ‘Retracting the past’). This is still a rare phenomenon: since 2015, the oldest retractions represent some 0.7% of the total number, up from 0.2% before 2010. But around 3% of expressions of concern have occurred for works older than 20 years. That illustrates how publishers might choose to issue expressions of concern for older papers, rather than retract, because it can be harder to investigate historical records.

What’s worthy of retraction?

In 2005, biochemist Barry Milborrow sought to retract five of his own papers, published between 1980 and 1982, on biochemical pathways that bacteria, algae and plants use to make pigments, because later work had disproved their findings. The Biochemical Journal, which published Milborrow’s self-retraction listing the five articles², retracted both articles it had published. Another journal issued a correction, but no notice was issued by the publisher of the remaining two papers. Ian Swift, Milborrow’s co-author and PhD student at the University of New South Wales in Sydney, Australia, at the time, says he only found out about the retractions many years later, after Milborrow had passed away.

According to COPE guidance, journal editors can retract articles if they no longer have confidence in the results and conclusions reported in them. By that rubric, many older papers could, in theory, be retracted, but, being rarely cited, are usually left alone.

Even when invalid papers are still being cited, retraction might not follow. In 1985, for instance, geneticist Svante Pääbo reported that he had cloned DNA from an Egyptian mummy³. The seminal work marked the birth of palaeogenomics, a field that earned Pääbo a Nobel prize in 2022. But the mummy DNA was actually contamination from a modern handler, as Pääbo, now at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, has openly acknowledged. “What is problematic is if false conclusions lead other scientists and the community as a whole astray,” he told Nature when asked for this article. “I don’t think anyone is currently misled by the mummy paper,” he says, which is why he argues that the work should stand, adding that to correct very old papers with errors “would create more confusion than clarity”.

Retractions can reshape scientists’ careers in unexpected ways

In practice, old papers tend to be considered for retraction only if they have widespread integrity issues that journal editors feel shouldn’t be allowed to stand unmarked. A major trigger for mass retractions of older work is when investigations of an individual’s research cast doubt over their entire body of work. For instance, a 2019 investigation by King’s College London into the work of psychologist Hans Eysenck and his German collaborator Ronald Grossarth-Maticek has led to papers dating back to the 1940s being flagged — albeit many with expressions of concern rather than retractions.

Eysenck died in 1997 as one of the most celebrated psychologists of the twentieth century. But in 2019, psychiatrist Anthony Pelosi, who had critiqued Eysenck’s work to no effect in the 1990s, published a renewed analysis of the studies⁴. An accompanying editorial called on King’s College London — where Eysenck worked from 1955 until 1983 — to conduct a formal investigation. The investigation report, also published in 2019, concluded that 24 articles by Eysenck and Maticek published between 1988 and 2000 were ‘unsafe’ and recommended they be retracted.

Those recommendations weren’t followed entirely. Only 16 of the 24 papers have been retracted, and another 4 have expressions of concern. Four remain unflagged. After sustained criticism over the narrow scope of the 2019 investigation, King’s College London published a statement in October 2025 saying that other publications based on the data set used in the 24 papers it examined are also likely to be ‘unsafe’ and should be flagged by journal editors — although it did not list affected publications.

So far, 94 of Eysenck’s hundreds of papers have been either retracted or tagged with an expression of concern, the oldest of which was published in 1943 (see ‘Longest gaps’).

Such inconsistencies in how publishers deal with research-integrity issues are common and don’t relate only to very old papers, says Andrew Grey, an endocrinologist and research-integrity specialist at the University of Auckland in New Zealand. He alerted publishers to concerns over large numbers of articles by Yoshihiro Sato, a Japanese bone-health researcher who fabricated data in dozens of clinical trials, and has reported seeing some journals respond with retractions or expressions of concern, whereas others take no action⁵.

One famous scandal of the past — the 1980s discrediting of work by John Darsee, a US biologist — has led to just 17 retractions. Although news reports at the time stated that more than 80 of Darsee’s papers had been retracted, this refers to editorial notes and warnings that, in parallel with the Gupta case, don’t appear in modern records as retractions.

Dangerous ideas

Another trigger for retracting old work is when it attracts renewed attention, particularly in medical fields.

Each year, Marcel van der Heyden, a physiologist at the University Medical Center Utrecht in the Netherlands, asks his final-year students to write a review article about a poison or toxin. In 2025, he chose caesium salts, and his students uncovered 20 case reports of people ingesting the toxic chemical⁶. Often, the individuals deliberately consumed the compound as an alternative treatment for cancer. Although the US Food and Drug Administration cautioned against its use in 2018, the deadly ‘therapy’ is still promoted in books and online by some wellness centres.

The treatment traces back to two papers published in 1984 in Pharmacology Biochemistry and Behavior. One of the papers was written by Austrian-born Hellfried Sartori, who had been prosecuted in Maryland in 1982 for fraud relating to using the unproven cancer treatment. That case ended in a mistrial. Sartori subsequently had his medical licence revoked in several US states in the 1980s and was sentenced to prison in Virginia in 1999 for practising medicine without a licence. And in 2010, an inquest into the deaths of five people in Australia found that he had been treating patients remotely from Thailand. Many more of his patients had also died under his treatment, the court heard.

The other paper promoting caesium therapy was by Wisconsin-based Aubrey Keith Brewer. Brewer died in 1986, but his namesake Brewer International Science Library continues to distribute information on caesium and other alternative therapies with a disclaimer that the protocols are “for information purposes only”, and that people should consult their physician if unwell. Nature contacted the website but received no response.

Van der Heyden contacted Pharmacology Biochemistry and Behavior in August 2025, suggesting an expression of concern could be warranted. “Without the original data, it’s hard to tell whether there is a mistake made, or whether it’s on purpose,” he says. But the research-integrity team at the journal’s publisher, Elsevier, decided in January this year to remove the two papers, soon after Van der Heyden informed them that his review had been accepted for publication and that his university was poised to issue a press release about it. The papers were deleted entirely this April, as Retraction Watch reported.

The move to scrub a paper, rather than merely retract and leave it in place, can be taken if the contents could present a serious health risk to the public, according to COPE guidelines.