Its been nearly a quarter of a century since McMahon and Greene suggested that the stiffness of running tracks could be adjusted to enhance a runner's performance. They demonstrated that tracks made to be compliant (deformed by the force of the runner's landing foot) could result in greater running speeds. Indeed, subsequent installation of several “tuned tracks” have resulted in increased running speeds as well as reduced running injuries. This early work was based on a surprisingly simple model of a running animal behaving like a single mass and spring. However, details of how surface stiffness relates to an animal's energetics and biomechanics have remained sketchy.
In this study, the authors set up a treadmill at Harvard that had one of five different platforms beneath the running area. These platforms resulted in a 12-fold range in track stiffness, from very stiff (≈950 kN m-1) to very compliant (≈75 kN m-1), including the range of track stiffnesses used in tuned tracks. By simultaneously measuring forces, kinematics (limb and body movements and angles) and oxygen uptake, the authors could test several hypotheses, including predictions that runners would make mechanical changes (in the knee in particular) to become “stiffer springs” while running on more compliant tracks, resulting in a lower cost of running.
What they found is that as the vertical displacement of the track increases (on increasingly compliant tracks), displacements in the runner's center of mass change very little, as predicted. However, sweep angles, stride frequencies, stride lengths and duty factor (time the foot is on the ground) are also nearly constant, independent of track stiffness. They had specifically postulated that these measures of leg posture would change to adjust leg stiffness inversely with track stiffness. Their results were somewhat surprising as the overall stiffness of the leg did indeed increase (by 29%), presumably by increasing the muscle stiffness. Further, as the track compliance increased, the metabolic cost of running decreased a great deal and did so nearly linearly, as elastic strain energy could be returned to the runner with each stride as the track deformation was restored. Further, because the “muscle machine” is not 100% efficient, for every watt of mechanical power returned by the track, the runner saves 1.8 W of metabolic power. The result is an overall reduction of 12% in the runner's metabolic rate on the most compliant track.
This paper is notable for at least two reasons. The first is the reminder that statistical significance is not sufficient to demonstrate biological significance. The authors rejected their “knee stiffness” hypothesis despite most results being statistically significant, as they argued convincingly that these results lacked the magnitude necessary not to reject the hypothesis. The second is that simple models should be retained as long as possible. While it is obvious that running is a complex behavior, the greatest insights often result from the simplest models.
What this study failed to do was find the “optimally compliant” running surface. As long as the resonant period of track plus runner is close to the surface contact times of the runners, an ideal track stiffness could be achieved. This experiment will remain for another day. In the meantime, those of us with dreams of setting our personal records for running distances best look to the most springy running surfaces we can find.
