Subscribe to Running Research News

Posted: November 4, 2005

Science of Sport: Any Limits To Running World Records?

By Owen Anderson, Ph. D. (Copyright © 2004-2005)

As you know, there has been a reduction in the frequency of world-record-breaking running performances in recent years. This drop-off has caused many exercise scientists to ask a basic question: "Are there limits to running world records?"

Expressed in this manner, the query is of course silly. After all, it is obvious that there are limits to running world records. No one will ever run a mile in faster than 3:20, a marathon in less than 1:58. A better question would be: "What are the specific performance limits beyond which human runners can not possibly progress?"

A number of different researchers have tackled this problem. A now-infamous study carried out by B. J. Whipp and S. A. Ward in the early 1990s looked at trends in world records over time and predicted that women marathon runners would run as fast as men in the year 1998, with a convergent time of 2:01.59. A key problem with this research was that it employed a simple linear regression model to forecast performances. Once a straight line is "fit" to the existing data (world-record running speeds as a function of time), there is of course a slope to the line and thus a continuous, upward movement in performances. In effect, there is no upper limit - runners will simply get faster and faster as the years roll by, according to the model.

An even-more-offbeat analysis was recently completed by A. J. Tatem and colleagues in which performance time, not running speed, was the "response variable" (the function of time). Again, linear regression was utilized in the analysis, and one of the predictions which emerged was that both male and female runners would be sprinting 100 meters in just 8.079 seconds by the year 2156. An interesting aspect of this kind of linear modeling, using performance time as the "y" variable, is that eventually runners will run "negative times" for all competitive distances. The world record would eventually be minus-.2 seconds for 100 meters, and so on.

Very recently, Alan Nevill and Gregory Whyte of the University of Wolverhampton and the English Institute of Sport have done the right thing: They have modeled middle- and long-distance running world records from the last 100 years using an S-shaped logistic curve, rather than a straight line. We won't go into the mathematical details of this model, but we will point out that the model assumes that world-record speeds increase slowly for awhile, then zoom upward at a high rate for some period of time (for various reasons, including better training techniques and broader participation by athletes from previously underrepresented parts of the world), and then settle back into a period of rare and modest increase (like the one now), followed by the attainment of an asymptote beyond which no further gains in world-record speed will be made.

As it turns out, the Nevill-Whyte model fits the world-record data from the last 100 years quite beautifully, statistically explaining most of the variation in performances at distances ranging from 800 meters all the way up to the marathon.

And since the model works so well, it can be used to predict the asymptotes - the predicted peak world records beyond which male and female runners will not progress.

Such predictions are interesting, to say the least. One is that the women's peak world record for 1500 meters has already been attained - at 3:50.46. Another is that the fastest-possible mile time for men is 3:39. For the marathon, the peak time is suggested to be 2:03.38.4. In addition, the race with the biggest projected improvement (~3.3 percent) in performance is the 5K, which is predicted to level off at about 12:12.1 for the peak world record. For most race distances, the projected improvement (before the peak is attained) is about 1 percent.

As your editor, I do have some problems with these numbers. For one thing, the model assumes that everything will stay the same in the future - that is, that there will be no unusual upgrades in the scientific understanding of performance and no extraordinary advancements in training techniques. On the training side of the equation, I actually think that enormous strides can be made. I also believe that as the globalization of sport continues new genotype-phenotype combinations will appear on the athletic stage and provide an additional boost to performance levels.

And - I do nurture the perhaps-too-irrational dreams that a special man will eventually finish a marathon in under two hours and an incredible woman will soar through 26.2 miles in faster than 2:11. When this happens, Nevill and Whyte will of course be astonished - and they will simply have to adjust their asymptotes!

As a final note, I have some great news for all of you: I have created a 5-K program which will help you smash through your personal 5-K asymptote.
To learn more about this, visit RunningResearchNews.com.

Copyright © 1998-2005 by Running Research News