I posted last week about a newly invented “cruise control” device for runners, which controls pace by cueing stride frequency with a metronome. That ignited an interesting discussion about how we change pace while running: do we take quicker strides, longer strides, or a combination of both?
For starters, Pete Larson send me a paper with some nice clear data that shows how the two factors interact (at least in one physically active group of 24 men and nine women between 18 and 34 years old):
The x-axis runs from 0 to 12 m/s, so the data runs from 2.5 to just over 9 m/s, which is 6:40/km to under 2:00/km — i.e. sprinting). Sure enough, stride length is a much bigger factor than stride frequency at typical jogging/running speeds, but the frequency curve is never perfectly flat.
I also exchanged a few e-mails with Max Donelan, one of the co-inventors of the cruise control, who explained a little more about how the device works and what sort of interactions between stride rate, length and running speed they saw in their testing. His answers were very interesting and well-explained, so with his permission I’m going to post them here rather than trying to summarize them. One of the most interesting points, I think, is that when his metronome cues runners to increase their stride rate, they also automatically increase stride length to arrive at the pace they’d naturally associate with the new cadence. Makes sense, but that hadn’t occurred to me.
Q: I’m also curious (as you saw in my blog entry) about how effective cadence is at controlling pace.
A: We have tested a number of subjects running at a range of speeds. It is absolutely true that some runners increase speed predominantly by increasing stride length. In fact, I would say that most runners that we have tested increase stride length more than frequency. However, all the runners we have tested also increase their frequency when they increase speed. We have yet to find a runner that only increases stride length or only increases stride frequency. For our purposes, it doesn’t matter whether people increase speed predominantly with increasing stride length as long as the relationship between speed and frequency is not perfectly flat (and we have yet to find a subject like that).
Of equal importance is a second phenomenon which is less intuitive. When someone is running at a particular cadence and you ask them to match a faster cadence, they not only increase their stride frequency but also their stride length.They alter both frequency and length to converge on the speed that they normally prefer at the new cadence. For example, a 10% increase in frequency might yield a 40% increase in length to get a 54% increase in speed. This allows us to use frequency to have control authority over speed.
Q: What sort of testing did you do?
A: We initially determined how frequency and length change with increases in speed by having subjects run at different steady state speeds on a treadmill. We carefully calibrated the treadmill speed and we measured step frequency with pressure sensitive foot switches. Stride length is simply speed divided by stride frequency.
To study how runners change both speed and step length when you give them an increase in cadence to match, we had them run overground with a metronome beeping in their ear. After a few minutes, the metronome frequency would rapidly increase to a new frequency. Subjects were instructed to match the beat. They were free to choose whatever speed they liked and, in principle, they could have stayed at the same speed. We measured step frequency with the same pressure sensitive foot switches. We measure and record overground running speed using a high-end GPS designed for quantifying acceleration in race cars.
We test our cruise control algorithm also during running overground. When we implement cruise control, we get runners within 0.5% of their desired average speed. This compares well with recreational athletes who average an 8% error, and collegiate runners who average a 4% error:
Green et al. Pacing accuracy in collegiate and recreational runners. Eur J Appl Physiol (2010) vol. 108 (3) pp. 567-572 http://dx.doi.org/10.1007/s00421-009-1257-5
For the recreational runners, an 8% error means that they will only be within 4 minutes of their target time for a 50 min 10 K. Running 4 minutes too fast may mean a surprisingly fast personal best, but it may also mean crashing and burning.
Very interesting stuff — both from a practical point of view (i.e. the cruise control), and for understanding more about how we run. Thanks to both Max and Pete for their contributions.