Foiling Fiddled Data

By Michael Edmonds 26/06/2013


Synthetic chemists produce new compounds by reacting various starting materials and/or reagents together. In order to determine the structure of what she has made, the chemist must run a series of analyses, which not only prove the compound has the structure she claims to have made, but also that it is present in pure form. These analyses include mass spectrometry, carbon-13 and proton nuclear magnetic resonance.

When a chemist submits a paper describing his work to a journal, the corresponding spectra for each compound accompanies the manuscript.

Synthetic chemistry and purification of compounds can be challenging and sometime additional peaks occur in the spectra due to difficult to remove solvents or impurities. And sometimes there is the temptation to remove these additional peaks using computer technology. And while such data manipulation does appear to be relatively rare, it has been enough to spur several chemistry journal editors to act.

In a recent article in Chemical and Engineering News, “Warning Shot on Data Integrity” *, Professor Amos B. Smith III, editor of the chemistry journal Organic Letters revealed that he has recently added a dedicated data analyst to his editorial staff to identify this sort of manipulation. He decided to do this after uncovering several cases where impurities appeared to have been “edited out” of spectra.

Professor Dale Poulter, editor of the Journal of Organic Chemistry, also makes use of a data analyst to check the veracity of spectra accompanying submitted manuscripts.

Other chemistry journals also have ways of detecting data manipulation. Angewandte Chemie doesn’t have dedicated data analysts but trained chemists are employed as editors, while at Nature Chemistry they make spot checks on figures in papers.

The removal of solvent peaks in spectra might sound to some to be a fairly minor offence, but it is not only dishonest, it can lead one into flawed ways of thinking. During my postdoc research I produced a compound which “fitted” what I expected to produce except for one extra peak in the carbon 13 NMR. It was so tempting to just ignore this peak, but after pondering the problem for days, I suddenly realised that it was not the compound I had expected but something far more interesting – something that opened a whole new area of research.

Honest presentation of spectra is always the best policy, and journal editors are now making sure that those submitting manuscripts appreciate this.

* There is no direct link currently available to this article in Chemical and Engineering News. The author of the article is Stephen K. Ritter.