by David Goodstein, Princeton University Press. Hardback ISBN 9780691139661, £15.95 ($22.95). E-book ISBN 9781400834570, $22.95.
Now that we can easily access scientific papers without leaving our offices, thanks to the availability of electronic versions of the most important journals, many of us at CERN rarely visit the library. Yet there are many other good reasons to stop by, among them the recently created bookshop, which houses a diverse collection of interesting books. This is where I first saw this book, in which David Goodstein shares his knowledge and reflections on scientific misconduct, including first-hand reports on some of the allegations that he studied as Caltech’s vice provost.
Throughout the book, Goodstein presents several cases with considerable detail, such that the reader is invited to judge whether scientific misconduct happened or not. The opening case describes Robert Millikan’s determination of the electron’s charge, based on measurements of 58 oil droplets, and addresses the allegation that this was a subset of all observations, selected because they were in line with the experimenter’s convictions. The verdict, “not guilty”, is supported by 22 informative pages, offering the reader a tour of the difficulties of the experiment – in the context of 1912 – including an explanation of why viscosity played a more important role than gravity or electricity in understanding the movement of the oil drops. I particularly enjoyed seeing four pages from the original notebooks where the measurements were written down. Rather than “manipulating” his data, Millikan carefully selected high-quality observations to obtain an accurate measurement: his result agrees with the modern value within its quoted 0.2% uncertainty.
The book also contains a highly entertaining report of the “strange and complex case of cold fusion”, following the saga from March 1989 to recent days. Here there is also no evidence of scientific fraud, defined by the author as “faking or fabricating data or plagiarism”. Martin Fleischmann and Stanley Pons should not have announced their “discovery” as they did (in a press conference) and when they did (too early, fearing to be scooped by someone else). Nuclear fusion on a tabletop would really be too good to be true and many physicists and electrochemists promptly dismissed those claims after finding that they could not reproduce the results in their own labs; but “self-delusion, misperceptions, unrealistic expectations and flawed experimentation are not instances of scientific fraud”. Real fraud, in physics, is illustrated by the putative discovery of element 118 by Victor Ninov (Lawrence Berkeley National Laboratory) and by the “remarkable” breakthroughs of Jan Hendrik Schön (Bell Labs) in the field of organic semiconductors. Caltech’s own problems of research misconduct are illustrated with two cases in biology.
The first chapter is particularly worth reading, reminding us of the main ideas of Francis Bacon and Karl Popper regarding the scientific method, although I prefer the elegant summary provided by Bo Anderson back in 1984: “Nature never tells you when you are right but only when you are wrong; therefore, you have only learned something when you disagree with the data.” Also, Richard Feynman argued that scientists should carefully report everything they are aware of that could invalidate their measurements or models. Goodstein shows that these laudable ideas are not really suitable in the real world, to ensure rapid and robust scientific progress. Based on his own experience, he argues that bad theories and the experiments that prove them wrong are “quickly and quietly forgotten”. Who has ever received a Nobel prize for showing that a model disagreed with data? After listing “fifteen seemingly plausible ethical principles for science”, he systematically reveals their insufficiencies as guidelines to sound scientific conduct and replaces them with a more pragmatic “user’s manual” on how to pursue a successful and honest career in science.
I would have liked to have seen more examples of scientific fraud, including cases of fabricated data in biology and medicine, but it is understandable that Goodstein prefers to address cases he knows well. Although a little “Caltech-centric”, this is an interesting and easy-to-read book, suitable for relaxing with at the end of the year.
