JEM's 2009 Tune-up

The Rockefeller University Press $30.00 J. Exp. Med. Vol. 206 No. 5 968-969 www.jem.org/cgi/doi/10.1084/jem.20090867 968 One mouse ≠ one experiment In the past several years, the editors of the JEM have noticed a troubling trend. Authors are increasingly submitting data for publication that derive from a single experiment. Whatever the driving force behind this trend, it is a worrying one. The independent verification of experimental data is essential to demonstrate the reproducibility of a result and is a fundamental tenet of scientific experimentation. Without independent replication, data lose rigor and publications lose credibility. At the heart of this issue is the definition of an independent experiment. The editors of the JEM define a single, independent experiment as one in which experimental and control groups (comprising individual mice, culture wells, etc.) are tested contemporaneously to answer a specific question. Each independent experiment must be repeated a sufficient number of times to demonstrate the reproducibility of the data. Authors have offered various justifications for submitting data from a single experiment. One argument that has repeatedly cropped up, particularly in the context of experiments involving bone marrow chimeras, is that one mouse equals one experiment. According to this reasoning, each mouse generates a new immune system from the transferred bone marrow, and the inevitable variability among individual animals renders each an experiment unto itself. What this argument fails to consider, however, is the fact that experiments performed on any given day could produce erroneous results for several reasons. For example, the bone marrow cells transferred into recipient mice could be contaminated. No matter how many chimeric mice are made or how many controls are performed, if they are tested contemporaneously, they constitute a single experiment. Another frequent justification for single experiments is that a sufficient number of mice were included in each experimental group to generate a statistically meaningful result. Although this is a laudable practice that should apply to all experiments, it has no bearing on the issue of experimental reproducibility. Others simply argue that certain experiments are too time consuming and/or expensive to justify repetition (on occasion authors have even claimed that repeated experiments would be prohibited by institutional animal care and use committees). As outlined in our revised Instructions to Authors, all figure legends must specify the number of times each experiment was independently performed, as well as the number of animals or replicates in each experimental group. Although we are reluctant to dictate a specific number of independent experiments that must be conducted in any given case, data used to support any conclusion of the study must be performed more than once and must be repeated a sufficient number of times to demonstrate reproducibility. The JEM continues to encourage submission of studies involving humans and nonhuman primates, and we understand that these studies cannot be readily repeated in their entirety. Vaccine studies, Heather L. Van Epps