document

© 1999 Macmillan Magazines Ltd

Until a few decades ago, fraud in sciencewas generally believed to be uncom-mon and, when it happened, it was

perceived as the action of the deranged. Sci-ence is supposedly self-correcting, and so thediscovery of fraud was seen as inevitable.However, as laboratory life has become morecompetitive, and especially where experi-ments are difficult to replicate, fraud andother types of serious misconduct havebecome less rare.

Often, according to those who have stud-ied the phenomenon, the motivation has notbeen to falsify but to cut corners in reaching aconclusion that the fraudster genuinelybelieves to be true. In the trusting world ofscience, such behaviour is hard to prevent.

But the growing problem of scientific mis-conduct brings with it increased disruption,more money and effort spent clearing up themess, and public distrust of science. So it hasbecome an increasing priority for researchagencies worldwide to find effective ways ofrooting out misconduct and, where appro-priate, dealing with it firmly. They are alsoseeking effective preventive measures — forexample, by encouraging adherence to clearprinciples of good scientific practice.

US research organizations have manyyears’ experience of attempting to deal withthe problem, which seems to be particularlyprevalent in biomedicine. Their Europeancounterparts have been slower off the mark,but are now starting to address the issue inearnest. In some cases this is the result ofdomestic experience; in others, an extensionof demands for responsibility and account-ability in all areas of public expenditure.

As well as exploiting the legwork that hasbeen done in the United States — for exam-ple, by using definitions of exactly what con-stitutes misconduct (see box, page 16) —European research institutions are trying toavoid the mistakes of the US systems, forexample in allowing investigations of allega-tions to drag on for years. They are also con-

centrating increasingly on finding ways toavoid fraud, for example by trying to removeboth the motivation and the opportunity.

In the process, the concept of self-regula-tion by the scientific community remainssacred on both sides of the Atlantic. “But theright to self-regulation is not sacred,” saysBruno Zimmermann, head of research affairsat the Deutsche Forschungsgemeinschaft,Germany’s main grants agency. “It must beearned by showing the public that we areensuring that research is being carried out infull honesty.”

How big is the problem? The extent of scientific misconduct of the‘fabrication-falsification-plagiarism’ type ishotly — and usually rather unscientifically —debated in both the United States and Europe.Some scientists fear that publicized cases aremerely the tip of an iceberg. Others remainconvinced that the overall incidence remainslow, and even those with direct experience ofmisconduct cases are often optimistic.

Eberhardt Hildt is the young postdoctoralfellow who blew the whistle on what appearsto be Europe’s biggest fraud case, where twoGerman cancer researchers published scoresof papers that include apparently fabricateddata. He says he still has faith in the funda-mental honesty of science. “Otherwise Iwould not have decided to stay in science.”Hildt now leads a research group at the Uni-versity of Munich.

But Imogen Evans, research strategy man-ager of the UK Medical Research Council,points out that “the evidence on misconductis impossible to interpret”. She suggests that“the question of how widespread scientificfraud might be is unanswerable”.

It is even difficult to gather accurate dataon the number of research misconduct casesthat have come to light, as funding agenciesand research organizations have differentprocedures for recording cases. At some USgovernment funding agencies, for example,the only way to learn about misconduct caseshidden in archives is to request documentsunder the Freedom of Information Act — atime-consuming process that can produce anincomplete picture, and require a court deci-sion for the release of critical details.

The Office of Research Integrity (ORI),which addresses misconduct in projectsfunded by the US National Institutes ofHealth (NIH), has the most comprehensivemethodology for reporting cases. It publishesregular newsletters that report the final deci-sions on cases, including the name of the way-ward researcher, his or her institution, andthe punishment meted out. This newslettercan also serve as the lone corrective device forpublished fabrications, as fears of lawsuitsmean that some journals baulk at retractingfaulty articles (see box, page 15).

The ORI received about 1,000 allegationsof misconduct between 1993 and 1997. Itclosed 150 investigations, deciding that mis-conduct had taken place in almost exactlyhalf. Of the 76 researchers found to have com-mitted misconduct, 71 per cent were barredfrom receiving federal research funds forbetween 18 months and eight years. Eightyper cent of cases involved the falsification ofresearch data. The total number of NIHgrants issued during the same five-year pe-riod was more than 150,000.

In Europe, only Scandinavian countrieshave analogous national committees thatcan collate such statistics, and all come upwith relatively low numbers. The DanishCommittee on Scientific Dishonesty was thefirst to be established, in 1992. Its responsi-bility was initially restricted to biomedicine,like the ORI, but its remit was extended tocover all sciences at the beginning of thisyear. The committee has considered only 25cases in its seven years of existence, andfound substance in only four of them.

The number of reported cases of miscon-duct is therefore relatively small compared tothe total volume of scientific publicationseach year. The Science Citation Index listednearly one million articles last year pub-lished in 5,600 selected journals — and this isprobably only one tenth of the total numberof scientific journals published worldwide.

But the few surveys that have been carriedout suggest that the level of misconduct may

briefing scientific misconduct

NATURE | VOL 398 | 4 MARCH 1999 | www.nature.com 13

Science comes toterms with thelessons of fraudAlthough the incidence of proven scientific fraud remains low, several high-profile cases have convinced the research community of the need for effectiveaction, in particular by enforcing codes of good laboratory practice.

On other pages |Fall-out from nuclear cover-ups: p14 |Publishing dilemmas: p15 | Defining fraud: p16

This Briefing was written by AlisonAbbott with additional reporting byRex Dalton and Asako Saegusa.

AD

RIA

N V

AN

DE

PLA

S


be significantly higher than that reported. Inone survey, published in American Scientistin 1993, for example, between six and nineper cent of respondents said they were per-sonally aware of results that had been plagia-rized or fabricated within their faculties.

A survey in 1995 of nearly 300 randomlyselected researchers in Norway revealed that22 per cent of respondents were aware of ‘seri-ous breaches of research ethical guidelines’,with nine per cent reporting that they had per-sonally contributed to such serious breaches.Nearly 60 per cent said they were aware of lessserious misconduct within their faculties. Yetthe Norwegian Committee on Scientific Dis-honesty has had only nine cases referred to itsince it was established in 1994 — only two ofwhich were found to have substance.

Whatever the true level of scientific mis-conduct, most agree that prevention is apriority. Kenneth J. Ryan chaired the Com-mission on Research Integrity whichreviewed US policy on scientific misconductin the mid-1990s at the request of Congress.He says that the commission’s remit “was nota question of punishment, which is whatmost scientists worry about, but of theintegrity of science — how to teach peopleand to prevent misconduct”.

This is certainly the prevailing view inEurope, where many scientists have watchedwith dismay how investigations of allegedfraud in the United States have becomeincreasingly protracted, demanding sub-

stantial amounts of time from the scien-tists who become involved.

Over-due process?Like most organizations in Europe —except Denmark — both the NIH andthe US National Science Foundation(NSF) rely on the academic institutionsthat receive their research funds toinvestigate allegations of misconduct inthe first instance, and then to report theirfindings to the funding agency. Agencyofficials then review the institution’s find-ings, investigate further if necessary, andrecommend appropriate sanctions.

Keeping the primary investigations atthe institutional level affords importantprotection to those accused. But it can alsocreate an enormous burden for the institu-tions themselves. An investigating commit-tee may spend many months analysing note-books containing raw data and interviewingwitnesses, a process than can be particularlytaxing in the United States when the accusedbrings in an aggressive defence lawyer.

So it can take years to reach a final decision.Pre-agreed time limits are frequently waivedby both parties when they become unrealistic.Lawyers for the accused may attack theintegrity of the individual who makes theaccusation, issue legal challenges to the inves-tigating committee members or universityadministrators, or accuse whistle-blowers ofinappropriate personal relationships.

A lawyer defending a researcher suspect-ed of misconduct in California once hired aprivate investigator to produce a psychologi-cal profile of the dean handling the investiga-tion, based on a letter he had written to theaccused, labelling the dean as paranoid.

Such tactics have persuaded some admin-istrators and scientists that they would neveragain participate in such investigations.William R. Brinkley, a dean at Baylor Collegeof Medicine in Houston, Texas, and presidentof the Federation of American Societies ofExperimental Biology (FASEB), has ob-served a long-running misconduct case at hisown institution (Nature 383, 107; 1996).“The toll it takes is enormous,” he says. “Peo-ple get wrapped up in these cases for years.They have told me they couldn’t afford to doit again; it destroys their careers.”

The experience is having a significantimpact on how fraud allegations are han-dled. Until recently, scientists, defence attor-neys and professional societies all stressedthe importance of ensuring due process forthe accused — a reaction to the fact that thishad not been afforded during some cases inthe 1980s. But many now question whetherthe pendulum has swung too far the otherway. They are asking whether accusedresearchers have been granted so much dueprocess that institutions are deterred frominvestigating allegations, so facilitating a cli-mate in which a corrupt scientist might gounpunished — or even unchallenged.

Even the current president of FASEB,which has long led the demand for maximumdue process for accused scientists, acknowl-edges that “the duplication of the [current]process wears everyone out”. “We may wellhave built in too much due process,” says


14 NATURE | VOL 398 | 4 MARCH 1999 | www.nature.com

Very few cases of scientific fraud have cometo light in Japan. Some attribute this toinstrinsic honesty among Japanesescientists, but others say it may merelyreflect a national culture in which peopleturn a blind eye to issues that could give riseto social tensions.

The hierarchical structure of universityfaculties, particularly within the old,conservative ‘imperial’ universities, and theautocratic powers of individual professors,certainly discourage scientists from blowingthe whistle on dubious scientific practices.

“Whistle-blowing is rare, and those boldenough to make accusations do not walk offunscathed — they may well lose their jobs,”says Masanori Kaji, a historian of science atTokyo Institute of Technology.

In addition, young scientists cannot hopeto influence suspect scientific practice insidethe laboratory. “Under the koza system atJapanese universities, where each researchgroup is led by a powerful professor, juniormembers of staff have almost no voice in theway research is carried out.”

As a government white paper (policydocument) on nuclear energy researchpointed out last year, however, public trustin science in general has plummeted in the

past few years following a series of nuclearaccidents and subsequent cover-ups,including fabrication of research data atnuclear fuel recycling plants.

The white paper said that peopleperceived scientists as “insular” and“unwilling to disclose or share details oftheir research due to preoccupation withsuccessful results”.

As a result of such incidents, scientistsare under increasing pressure to take theethical aspects of research more seriously.Nearly half of the respondents in a recentpublic survey conducted by KanazawaInstitute of Technology said they did notbelieve that scientists have higher moralvalues than the rest of the population.Ninety-five per cent thought scientistsshould take more responsibility for the waythey work and how they communicate thedetails of their research to the public.

So, despite the apparently low incidenceof scientific misconduct in Japan, broaderscepticism towards ethical conduct inresearch is relatively high. Partly in response,the Science Council of Japan — a governmentadvisory group — published a report lastyear calling for a new system to promotegood scientific practice at universities. A. S.

Japanese scandals raise public distrust

Straight and narrow: the Scandinavians wereamong the first to develop research guidelines.

AN

TJE

LA

NG

ER


Brinkley. “It deserves a careful look.” Hehopes that FASEB’s new Committee on Sci-ence Policy will be able to review such issues.The committee’s chairman, David Brautigan,agrees. Due process may be abused, he says.“Litigation has become the defence.”

Europe looks for a better way The European research community has so faravoided committing itself to such lengthyprocedures — and hopes to keep things thatway. In drawing up rules and procedures forhandling scientific misconduct, Europeanresearch organizations have tried to ensurethat investigations are concluded rapidly.

In principle, the procedures, based on‘peer’ investigation committees and a com-mitment to protect both the whistle-blowerand the accused, are not radically differentfrom those in the United States. But mostEuropean organizations try to ensure speedyresolution of investigations by limiting thenumber of procedural stages.

Rules introduced last year at the Univer-sity of Freiburg, Germany, for example,require an informal investigation into thevalidity of a suspicion of dishonesty, fol-lowed, if necessary, by a formal investigation.The findings of the formal committee arepassed on, with appropriate recommenda-tions for sanctions, to the university presi-dent, who then decides on sanctions.

Albin Eser, a professor of law at the univer-sity and a director of the Max Planck Institutefor Foreign and International Criminal Law,helped draw up these rules and those of theMax Planck Society, which are similarly suc-cinct. “We deliberately avoid an appeal stage,because it would extend the time needed toreach a resolution,” he says. “In any case, asanctioned scientist can always appealthrough the courts.”

Scandinavian countries also leave appealsto the law courts. In contrast, the UK MedicalResearch Council, whose procedures are oth-erwise similar to those in Germany, insists on

incorporating its own appeals procedure, onthe grounds of fairness.

National versus localThe issues of whether countries should set upnational committees to investigate allega-tions of scientific misconduct, and whethersuch bodies, rather than local institutions,should be the first line for investigations, arevery sensitive in Europe.

Denmark is the only country that entruststhe investigation of all allegations of scientificmisconduct in the first instance to a nationalbody, the Committee on Scientific Dishon-esty, which is chaired by a high court judge.Daniel Andersen, its vice-chairman, stronglydefends the centralization of investigations:“We discourage cases from being handledlocally as there is a natural reluctance on thepart of a university to label one of its scientists— particularly a prominent faculty member— as a cheat.”

Claude Griscelli, director general of the



Many journal editors believe that breaches ofthe traditional ethics of scientific publicationare increasing. But few are confident of howthey should react. Should they retract apaper which only some of its authors, or acommittee investigating alleged misconduct,say includes fraudulent data? Should they‘blow the whistle’ on the authors of suspectsubmissions that they intend to reject?

There has recently been some movementon these issues. In Britain, for example, agroup of particularly frustrated medicaljournal editors recently created theCommittee on Publication Ethics (COPE) asa sort of informal self-help group. Separately,some individual journal editors are startingto develop their own policies.

Many journals, including all thosepublished by the US National Institutes ofHealth, now require each author to sign astatement accepting responsibility for thewhole content of a paper bearing his or hername — a move intended to avoid the perilsof honorary authorship.

One significant practical test of thewillingness of journals to help contain theimpact of fraudulent publication is a projectset up last year by the DeutscheForschungsgemeinschaft and the MildredScheel Foundation, and headed by Ulf Rapp,professor of molecular and cell biology at theUniversity of Würzburg.

The project will systematically determinehow many of the 550 journal papers and 80-odd book chapters written by two Germancancer researchers, Friedhelm Herrmannand Marion Brach, and some of their formercolleagues, included apparently fabricateddata. Local investigation committees hadalready identified 47 papers which appeared

to include fraudulent data. The task force,now one-fifth of the way through its stack,has identified a further 11.

Last June, Rapp approached 120 editorsand publishers to request copies of relevantarticles, copyright assignments andreviewers’ comments. His experience, hesays, was “rather depressing”. Out of 70replies, 40 sent copies of the articles, but onlyfour provided copyright assignments, andonly five reviewers’ comments.

Some editors, although keen to help, didnot hold files for long after publication, hesaid. But most refused to provide reviewers’reports, even anonymously, on the groundsof confidentiality. “Some seemed to considerus impertinent disturbers of the peace.”

More than half of the 18 journalsidentified by the original investigationcommittees as likely to have publishedpapers including fabricated data have notissued retractions. One is Blood, whichpublished seven of the papers. Its editor,Kenneth Kaushansky, professor ofhaematology at the University ofWashington School of Medicine in Seattle,says he is “uncomfortable” with this, but hadbeen advised against retraction by Blood’snew advisory committee on scientificintegrity, on the grounds that “definitive”proof of fabrication had not been provided.

The issue raises a general dilemma foreditors: if authors themselves are not willingor available to retract a paper, whosejudgement that a paper includes fraudulentdata should a journal accept in order to makethe decision itself?

Of those editors who publishedretractions of the data involved in theHerrmann and Brach affair, some acted on

the requests of co-authors, others oninformation from one of the localinvestigation committees. But others remainuncertain how to react. “We need a Europeanconference of editors to help us defineexactly how, and on what basis, to retract,”says Nicole Muller-Bérat, the Paris-basededitor-in-chief of Leukemia.

Nature’s policy is to publish whateverinformation it can about published papersthat have proved suspect or false, says editorPhilip Campbell, “although only after dueconsultation. Where appropriate we try topersuade the authors to issue a formalretraction, as promptly as possible. But thereare times when authors will not agree amongthemselves. We cannot act as a judge or jury,but we can alert the community to thesituation, and publish a statement by someof the authors while also making it clearwhere there is a disagreement.”

But defining policies on retractingpublished papers is a relatively simple taskfor journal editors, compared to that ofdefining policies on what to do whensuspicions of misconduct are raised in theminds of reviewers before publication.

Some journals, such as the Journal ofImmunology, are willing to informinstitutions of suspicions raised by papers ifthe authors do not give an acceptableexplanation. But many editors are notconvinced that their role should stretch so far.

“Journals should not act like secretpolice,” says Magne Nylenna, editor of theJournal of the Norwegian MedicalAssociation. “We consider submissions withthe utmost confidentiality, and it is not clearthat we should go so far as to informinstitutes of [suspected] wrongdoing.”

Editors debate whether to blow the whistle on suspect papers


French medical research organizationINSERM, would like to go further. He sug-gests that, at least when a local institution feelsunable to handle a particular, and potentiallyserious, case, only an INSERM committeewith members drawn from other Europeancountries could come up with the necessaryexpertise and distance from local academicpolitics to be able to carry out a truly indepen-dent investigation. The European ScienceFoundation is considering whether such acommittee should be set up under its aegis.

In contrast, even other Scandinavianresearch organizations that have adopted theDanish model, including Norway, Swedenand Finland, do not require institutions torefer cases in the first instance to the nationalcommittee. Local institutions can choose toconduct preliminary investigations.

In Norway, there were strong objectionseven to the concept of a national committee.Jarla Ofstad, a professor of medicine at theUniversity of Bergen, and a former chairmanof the Norwegian Medical Ethics Committee,argued strongly against the creation of theNorwegian Committee on Scientific Dishon-esty because it would “create too muchbureaucracy around an infrequent problem”.Including a judge in each of the Scandinaviancommittees turns them into “amateur court-rooms”, he says.

The Committee for Publication Ethics, aninformal group of British medical journaleditors, has long campaigned for a UKnational committee to handle all incidencesof misconduct, at least in medical research.This proposal is currently being discussed bya committee set up by the General MedicalCouncil, which sets professional standards,including research standards, for clinicians.

In Germany, however, where there isstrong distrust of any institution with cen-tralized power, opposition to a national bodyto investigate misconduct remains strong.

German complacency endedGermany is the most advanced of the non-Scandinavian European countries in insti-tuting procedures for handling and prevent-

ing scientific misconduct. Procedures forhandling allegations were drawn up by theMax Planck Society in 1997, only monthsbefore the first of a series of major fraudscandals shattered German complacencythat the country was culturally immune towhat had been seen as an ‘American scourge’.

The Deutsche Forschungsgemeinschaft(DFG), Germany’s main grants agency, wasdeeply shocked by the first case, when localinvestigation committees announced in 1997their “strong suspicions” that 47 papers bytwo high-flying cancer researchers, Fried-helm Herrmann and Marion Brach, had

included fabricated data (see Nature 387,750; 1997).

Former DFG president Wolfgang Früh-wald had said with some pride in 1995 that“the incentive [for a researcher] to falsify datato accelerate his career is greater in [the US]system than in the German research system,which is tightly controlled by self-regula-tion”. Almost as a gesture of atonement forthis naive assumption, he ensured that thefirst guidelines for good scientific practice inGermany were rapidly developed before histerm of office expired at the end of 1997.

These guidelines are intended to tightenup self-regulation rather than provide a sub-stitute for it. Their publication promptedaction by research organizations elsewhere inEurope, which had previously been consid-ering in a more leisurely manner how todevelop procedures for handling — and pre-venting — scientific misconduct.

In Britain, the Biotechnology and Biolog-ical Sciences Research Council launchedgood practice guidelines at the end of lastyear, and other research councils are follow-ing suit. In France, INSERM distributedguidelines to its institutions last month, andthe biggest French research organization,CNRS, is currently debating the issue.

There is little to differentiate the goodpractice guidelines that have been developedwithin different research organizations. All


16 NATURE | VOL 398 | 4 MARCH 1999 | www.nature.com

The US Office of Research Integrity definesmisconduct as “fabrication, falsification,plagiarism [‘FFP’], or other practices thatseriously deviate from those that arecommonly accepted within the scientificcommunity for proposing, conducting, orreporting research. It does not includehonest error or honest differences ininterpretations or judgements of data”.

Most US organizations have adopted thisdefinition, with minor amendments oradditions. Three years ago, however, theCommission on Research Integrity — whichhad been set up by Congress to, among otherthings, refine the so-called ‘FFP’ standard —proposed that there should be a more precisedefinition.

The commission suggested that a newdefinition would help the courts to decidecomplex cases. One suggestion was that theterm ‘elimination’ should be included,referring to the deliberate omission of dataconsidered inconvenient in ensuring thedesired result.

But the proposed changes haveencountered considerable hostility fromscientists. Many fear, for example, thatidentifying such a practice as misconductcould lead to sloppy record-keepingbecoming grounds for career-damagingmisconduct allegations.

The recommendations are currentlybeing tossed around the White House’sOffice of Science and Technology Policy, andthis has given scientists another reason toworry. This is because any new definition toemerge from this office may applygovernment-wide, covering research fundedby the National Institutes of Health, theNational Science Foundation, and manyother agencies involving energy, defence,education and veterans’ hospitals.

Kenneth J. Ryan, a Harvard Universityphysician and professor emeritus whochaired the Commission on ResearchIntegrity, says that policymakers are awareof the hostility surrounding the proposals,and have therefore been in no hurry toconclude their deliberations, which havenow stretched over years.

Some federal officials working on thenew policies say that the slow pace is to beexpected, given the importance of the issue.“It did, after all, take nearly a decade toformulate new regulations for humansubjects research,” says one official.

But scientists and officials involved in thework of the commission fear the importanceof the issue has “dropped off the radarscreen”, and that it will stay there until anegregious case reaches a high enough profileto prod policymakers into action. R. D.

US stalls on new definitions of misconduct

Smoking gun: investigators in the Herrmann and Brach case electronically separated the componentsof this autoradiogram and found it to contain duplicate data. Lane 1 (0 hrs) is the same as lane 4 (48hrs), and lane 2 is the same data as lane 3, but part of the figure is rotated.

J. I

MM

UN

OL.


stress the importance ofissues such as establish-ing clear lines of respon-sibility in a researchteam, of conductingexperiments in an openway, of good lab note-books and archiving ofdata, the value of men-tors and the teaching ofgood practice to youngresearchers, and re-sponsibility in pub-lishing.

In addition, the DFGguidelines recommend the appointment ofan institution ombudsman, a sort of official‘uncle’ with whom young researchers coulddiscuss in confidence their concerns aboutdishonesty in their laboratories. The Her-rmann and Brach affair had revealed howdangerous the hierarchical culture of a Ger-man laboratory can be, since most young sci-entists are afraid to challenge the all-powerfulfigure of the professor.

But, while most researchers accept thatcodes of good practice are important in rais-ing awareness, they also accept that suchcodes cannot in themselves eliminate mis-conduct: the culture of a particular laborato-ry, and its individuals, can always connive toignore or undermine them.

According to Eberhardt Hildt, the whis-tle-blower in the Herrmann and Brach case, itwould have helped if there had been anombudsman at the Max Delbrück Centre forMolecular Medicine in Berlin, where heworked with the two researchers. It wasHildt’s PhD supervisor in distant Munichwho acted in this capacity and helped him tobring the affair to the attention of a disbeliev-ing community. But, he says, good notebookkeeping would not have helped.

In contrast, one young PhD student, whodeclines to be identified for fear of retribu-tion, says that an ombudsman would nothave helped him to expose the wrongdoing hewitnessed. He was nearly driven out ofresearch by his undergraduate experience ofregular, and unexposed, manipulation ofdata by group leaders in a big university labo-ratory in Germany where he conducted hisfinal-year project.

Young scientists in the laboratory wherethe young PhD student witnessed miscon-duct judged the risk too high to make itworth exposing what they saw as “minormisconduct” — the exaggeration of thenumber of controls, or the omission of datapoints to clean up a graph, to speed the pathto publication.

He says the laboratory chief, a successfulbut demanding task master, is almost certain-ly unaware that his group leaders are massag-ing data, and conducting experiments thatthey do not report at lab meetings, in order tohave results to show him in a dry period. “For

the group leaders it is notfraud or deviation fromgood practice: it is good tac-tics,” says the student.

Enforcing good practice The DFG predicted that mostuniversities, which guard theirindependence closely, wouldneed a serious incentive tointroduce procedures for han-dling fraud allegations andencouraging good scientificpractice. So it states in its own guidelines thatinstitutes which do not have rules by 2002will not be eligible to receive its funds.

The threat seems to have worked: ten outof 82 universities have brought out guidelinesand, according to Josef Langer, general secre-tary of the German Universities Rectors Con-ference, the others have all agreed to do so.

Britain’s research councils have developeda less direct way of persuading universities toadopt codes of practice. By requiring grantholders to sign a statement that the work willbe carried out in institutions with acceptablecodes of practice, they transfer responsibilityto the grant holder if misconduct occurs.

So far French universities are not beingpressured to adopt codes of practice. Aspokesman for the French University Presi-dents Conference insists that such codes arenot necessary “as university science is intrinsi-cally honest”. A similar situation exists in Italy.

US research organizations operate anassurance system in grant giving, similar inprinciple to the new system of the UKresearch councils. To be eligible for grants,institutions must agree to their researchbeing conducted and monitored according toa long list of conditions, which include thesubstance of the good practice codes beingdiscussed in Europe. The enthusiasm withwhich good practice is enforced varies greatlybetween institutions, however. The NIHencourages awareness by offering traininggrants for young scientists with the stipula-tion that they must be taught research ethics.

Reparations Most institutes try to ensure that papers areretracted when a scientific misconduct case isproven. But research organizations in Ger-

many in particular have made it a priority todevote substantial energy to isolating andrepairing damage to the research communityincurred by scientific fraud.

To disentangle truth from lies in the Herr-mann and Brach affair, for example, the DFGand the Mildred Scheel Foundation —which had both given grants to the tworesearchers — set up an investigation to tracethe history of the fabricated data, and deter-mine if the alleged practice of fabrication hadspread to other institutes from Herrmannand Brach. The investigators used software toanalyse the components of data presented inthe 550 papers and 80 book chapters writtenby Herrmann and Brach, and some formercolleagues (see figure opposite). It will take 18months to complete, and the results, detailingthe corrected record, will be published in aninternational journal.

In a similar fashion, the Max Planck Soci-ety required its Institute for Plant Breeding inCologne to repeat all the experiments report-ed in papers that had relied on assays that atechnician later admitted she had manipulat-ed (see Nature 393, 293; 1998).

This task was completed within a fewmonths with the help of researchers fromwithin the institute and from outside. Theresults, and the list of papers that containeddata that cannot be reproduced, are beingpublished this month in the Plant Journal.

Research organizations in other Euro-pean countries, as well as in the United States,have expressed admiration for this commit-ment to damage limitation. But none has, asyet, decided that it should do the same.



Hildt: favours ideaof ombudsmen.

The following websites include furtherdetails of guidelines on good scientificpractice or guidelines for handlingallegations of scientific misconduct,in English.● US Office of Research Integrity ori.dhhs.gov/regguide.htm● UK Medical Research Councilwww.mrc.ac.uk/mis_con.pdfor www.mrc.ac.uk/w_n1.html

● UK Biotechnology and Biological SciencesResearch Councilwww.bbsrc.ac.uk/opennet/structur/hrg/sciconco.html● Germany’s Max Planck Societywww.mpg.de/fehlengl.htm● Deutsche Forschungsgemeinschaftwww.dfg.de/aktuell/self_regulation.htm● Danish Committee on Scientific Dishonestywww.forskraad.dk/spec-udv/uvvu/

Guidelines on the web

Nota bene: accurate lab records can be

a defence against charges of fraud.

AN

TJE

LA

NG

ER

AN

TJE

LA

NG

ER

Documents

document