Sir

Further to T. Häusler's 'In Retrospect' review of Sinclair Lewis's 1925 novel Arrowsmith (Nature 453, 38; 2008), the book was required reading for graduate courses in professional practice and ethics in the biological sciences that I taught in the 1990s. Arrowsmith's ethical dilemma was whether he should deny some villagers his phage therapy so that they could serve as controls. His conundrum endures to this day — the choice between bequeathing knowledge from a properly designed controlled experiment and risking the health of members of the control group by withholding potentially beneficial treatment.

The control group provides an unbiased test of the null hypothesis, which predicts what to expect if our ideas of how nature works are wrong. It could be argued that it is therefore an ethical obligation for the scientist to take the null hypothesis seriously. No other professional is ethically obliged to consider what might happen if he or she is wrong.

Arrowsmith's employers put the entire ethical burden of choosing the control group and implementing the experiment onto his shoulders. Ethical burdens, however, are properly borne by the entire community. Current practices of having institutional review boards to oversee experiments, obtaining consent from the treated patient, double-blind procedures and full disclosure combine to ensure that the ethical burden of research is shared by all the people involved in the work and is not unfairly placed on a single individual.