Sir

Your Special Report (Nature 435, 258–259; 2005) on plagiarism in scientific texts overlooks an important related problem, namely multiple publications of the same data (graphs or pictures).

The computer programs for identifying duplicate text in manuscripts do not seem capable of handling this particular kind of dishonesty. In one case I have identified, the authors had published the same electron micrographs repeatedly in different journals, and attempted to do so with a manuscript that was sent to me for review. In trying to hide self-plagiarism, the copied figures were presented upside down, rotated by 90°, or slightly cut or expanded. This is reminiscent of Jan Hendrik Schön's scientific misconduct (Nature 417, 367–368; 2002).

When I informed the editor, she rejected the paper with a letter stating that it was unacceptable, as a reviewer had detected the plagiarism. But there are good arguments for journals to go further and take punitive measures, such as notifying the authors' institution and, if applicable, their funding agency. Papers identified as fraudulent after publication should be removed from the web version of the journal and be replaced by a note disclosing the scientific misconduct. The authors' consent should not be required.

Even more serious sanctions may need to be discussed publicly and included in the guidelines for manuscript submission. For example, the authors could be banned from publishing, permanently or temporarily, at least in the cheated journal. Funding agencies could add pressure by stating that scientists practising scientific misconduct will be denied the right to submit proposals.

Last, but not least, many cases of scientific misconduct could be avoided if the work of good reviewers were appreciated more explicitly. I tell myself that I'm doing it for the benefit of science. But, lacking appreciation, some reviewers might do their job in a sloppy manner. Why not encourage them? The only journal I know that makes an award for excellence of review is Environmental Science and Technology. Publishing a list of the top 10 or 20 reviewers each year may encourage more in-depth evaluations of manuscripts.