No matter what the terrain, we usually adjust our gait to stagger on. Even if it means adopting some rather strange styles, we adjust speed, step length and stride frequency to overcome most obstacles. John Bertram began investigating the mechanics of walking several years ago. Controlling stride length, Bertram was surprised to realise that the athletes he was investigating adapted spontaneously to relatively unnatural gaits, adjusting their speed and frequency in unexpected ways. `If we'd seen this in animals'says Bertram' we'd have thought we'd interfered with them'. But the athletes seemed completely unaware of their adjustment. Bertram realised that the walking gait is very plastic and that he could predict these strange behaviours by minimising the energetic cost of locomotion. But was this optimisation an artefact of walking, or a general characteristic of human locomotion? Bertram needed to investigate another gait, running(p. 622).

Based at Florida State University, Bertram had no trouble finding outstanding athletes to volunteer for running practice and a convenient track to test them on. Having found 5 willing participants, Bertram, Brian Jacobi and Michael Butcher set the athletes three tasks; running at set speeds on a treadmill, running at set frequencies striding in time to a metronome, and running with fixed stride lengths by stepping on markers on a field. Needing spontaneous responses to the running conditions, Bertram only allowed the runners two practice runs to find their rhythm, before recording the performance and monitoring how the runners adjusted to the constrained conditions.

Just as Bertram had found with the walkers, the Florida runners adjusted naturally to the unusual running styles and came up with some unexpected, but natural, solutions. Bertram realised that speed, stride frequency and length were linked; adjusting one forced the runner to modify the other two to compensate. The runners must be adapting to optimise some aspect of their performance. Bertram turned to the metabolic cost.

But measuring the volunteer's metabolic costs as they repeated the constrained running tasks was going to be too tricky. Bertram realised that the metabolic data might already be out there in the literature. At this point Anne Gutmann, a Cornell graduate student visiting Bertram's lab for a year,joined the team. She began trawling through journals, and identified 4 well documented running studies that she could extract reliable, constrained metabolic cost data from. Using sophisticated mathematics Gutmann constructed a `cost of transport' surface as a function of the runners' speed, step length and running frequency, and used this to predict how a runner's gait would respond if any of these parameters were constrained. Gutmann also compared how Bertram's runners faired on the Florida running track with the metabolic data she'd extracted from the literature. The running behaviours agreed remarkably well, although the behaviours weren't completely identical.

So, constrained optimisation of the metabolic cost of movement allows the prediction of both walking and running gaits because we always get by for the least possible cost. And the relationship between our speed, stride length and frequency as we walk or run `is not an accident of the mechanics; the body is monitoring gait all the time' says Bertram.

Gutmann, A. K., Jacobi, B., Butcher, M. T. and Bertram, J. E. A. (
). Constrained optimization in human running.
J. Exp. Biol.