X-rays, penicillin and the World Wide Web — these are often-cited examples of how unshackled science can lead curious researchers to unforeseeable discoveries. But such discoveries quickly get taken for granted, so it is good occasionally to highlight new examples of the unexpected.

One can be found in research into fitness and exercise. Few if any people could have anticipated the results that poured out of the long-term, US-based HERITAGE Family Study into exercise training. HERITAGE represents research at its most basic: engage large numbers of people, make them do fitness training for 20 weeks, gather diverse physiological, biochemical and genetic data with no hypothesis in mind, and see what you get.

The researchers found an astonishing variability in our individual responses to training (see page 188). Some participants got much fitter, others hardly at all. And the study shook assumptions about the presumed risk factors for common diseases that are particularly prevalent among couch potatoes, such as cardiovascular disease and type 2 diabetes. In some outliers the overall profile of presumed risk factors was dramatically high, for others quite small.

Meanwhile, a UK-based study on athletes has led directly to clinical trials of a drug targeted to a particular ‘fitness gene’ in highly vulnerable groups of patients — premature babies, for example — whose very life depends on their efficient use of oxygen, just like an athlete's ability to win medals.

Scholars of science, policy-makers and those who simply enjoy the highways and byways of research, take note.