Andrew Maynard rightly points out that 'one-size-fits-all' definitions are potentially harmful as they cannot capture the range of risks associated with a diversity of nanomaterials (Nature 475, 31; 2011). But in arguing that regulators should instead work with a list of functional attributes to develop proper science-based regulations, he overlooks the fact that knowledge of such attributes, and internationally agreed methods for acquiring it, will take years of research to accumulate.

Given this situation, adopting a functional definition arguably means that, in theory, the most scientifically sound approach to regulation would be to impose a moratorium on the use of nanomaterials until the relevant trigger-point attributes can be established. There would still be the problem of defining the point in the life cycle of a nanoproduct at which such attributes should be measured, given that they would be likely to change with time and context.

The concept of adaptive regulation itself needs to be flexible enough to recognize the limitations of scientific understanding and its legitimacy as a basis for policy.