The credibility of science now rests with co-authors, who need to ensure that they check a paper's validity before it is published, says Robert Park.
It has been the summer of lost faith. Each day has brought news of another Wall Street scandal, as millions of investors see their savings vanish, stolen by corporate executives who have conspired with auditing firms to conceal huge losses. And each day, more Catholic priests in the US have been charged with sexually abusing children. But surely we can trust scientists, can't we? Maybe not.
Bell Laboratories, the world's most renowned industrial laboratory, announced an investigation into charges that physicist Jan Hendrik Schon, a miracle worker in the exciting field of nanoelectronics, had fabricated research results. Schon had been showered with honours for a series of remarkable papers involving the use of single organic molecules as electronic switches, but researchers at other laboratories were unable to repeat Schon's experiments.
Lucent Technologies, the parent company of Bell Labs, appointed a panel of distinguished scientists from outside the lab to investigate. This unprecedented action reflects the importance Lucent attaches to the almost mythic reputation of Bell Labs.
These are difficult experiments, and it would not be unheard of for other scientists to have difficulty repeating them. But then the investigation was expanded to include four papers on superconductivity. Charges of fabricated results in two very different areas of research make it much less likely that an innocent explanation will be found.
While physicists were still buzzing about the scandal at Bell Labs, the Lawrence Berkeley National Laboratory fired Victor Ninov, the physicist who had been credited in 1999 with the discovery of elements 118 and 116. The discovery had been hailed as a "beautiful" confirmation of a predicted "island of stability" of super-heavy elements. Scientists at other labs, however, were unable to confirm Ninov's experimental findings. After a year-long internal investigation, the laboratory concluded that Ninov had fabricated the results.
This was not supposed to happen in physics. Not because physicists have any greater claim to wisdom or virtue, but because the credibility of science is anchored in the willingness of scientists to expose ideas and results to independent testing and replication by other scientists. Nothing is supposed to be held back. Perhaps in the softer sciences, where judgements are more subjective, fabricated results might survive for a time, but not in a rigorous, quantitative science such as physics. The inevitability of exposure, it was thought, would deter misconduct.
Some physicists could not resist crowing that these cases prove that the system is working. Both labs responded to charges of misconduct with thorough investigations. But questions have been raised about earlier research by the two physicists at laboratories in Europe. If fabricated results turn up in a re-examination of earlier work, the boast that the system is working won't fly.
Should the journals have spotted problems earlier? There are warning signs editors look for to eliminate crackpot papers, but none of them applies here. These were experienced researchers, at distinguished laboratories, following all the rules of scientific openness. But the peer reviewer has no way of knowing if the scientific apparatus is working properly or if the researcher is honestly recording the readings.
However, no one seems to be talking about the most serious question. Where were the co-authors? These scientists did not work alone. On the paper in question, Ninov had 14 co-authors. Fourteen scientists who, in allowing their names to go on the paper, certified its validity. Schon's co-authors included the person to whom he answered, Bertram Batlogg, a highly respected scientist who shared many of the honours with Schon. Automatic co-authorship of a person who has not made a substantive contribution is forbidden by every code of professional conduct. He must have been in a position to know what was going on.
The lesson from this summer of lost faith is that scientists must satisfy themselves of a paper's validity before allowing their names to be added as co-authors. The success and credibility of science depend on it.
Robert L. Park is professor of physics at the University of Maryland and author of Voodoo Science: The Road from Foolishness to Fraud , published by Oxford University Press, £8.99.