166 Brit.J.Sports Med.- Vol. 21, No. 4, December 1987, pp. 166-167 GROUPING OF RUNNERS DURING MARATHON COMPETITION K. YAMAJI, PhD and R. J. SHEPHARD, MD, PhD Br J Sports Med: first published as 10.1136/bjsm.21.4.166 on 1 December 1987. Downloaded from Laboratory for Exercise Physiology, Faculty ofEducation, Toyama University, Toyama, Japan and School of Physical and Health Education, University of Toronto ABSTRACT Marathon runners tend to form clusters during competition, possibly as a means of improving their performance. Studies of international competitors in Fukuoka and Tokyo races have shown a density of 0.44 to 0.80 runners.m2 among those maintaining a pace of 4.8-5.6 m.s-1, clustering being most marked among elite performers. Some 95% of clustered runners maintained a minimum distance of 0.5-1.5 m from other competitors; moreover, such competitors avoided occupying an angle of ± 15 degrees either before or behind other runners. The phenomenon merits closer study by those interested in the bettering of marathon performance. INTRODUCTION The video records were obtained as subjects ran away Today's marathon races have been called "survival events". from the cameras. The inter-personal distance and angle Typically, the top contestants form a cluster immediately were estimated by standard geometric principles, after the start of the race and victory goes to the survivor correcting for the height of the camera at each of the five among this elite group. Individual runners apparently adjust selected sites and the distance of individual runners from their pace quite finely so that they can sustain this the building where the camera was mounted. Distances clustering. between groups and between individual runners were calculated from the passing time of each competitor. For In order to study the phenomenon in detail, we mounted example, at the 9.8 km point in Fukuoka, the majority of a video-camera on a high building at two points along the competitors had grouped themselves into eight clusters course of the 1985 Fukuoka International Marathon Race (at (Fig. 1). Where there were more than 6 runners in a cluster 9.8 and 20.0 km), and at three points along the 1986 Tokyo the density of the group was also calculated (runners.m2), International Marathon Race (at 4.8, 16.0 and 24.5 km). based on the distance from the first to the last member of METHODS the cluster. A rectangle was created enclosing all members of the cluster, and this was arbitrarily divided into two Although the Fukuoka and Tokyo races attracted a fairly segments (high density and low density); the mean density small number of competitors (137 and 68 respectively), they was then calculated as the average of densities for the two were international events with a fairly high level of segments. Where the interpersonal distance was more than performance, as can be judged from the pace at the 9.8 km five metres, a competitor was considered as running alone. mark in the Fukuoka race (4.8 to 5.6 m.s-1). Fukuoka Marathon (1985) at 9.8 km point http://bjsm.bmj.com/ 0 0 5 10im I~ l0in 0 * 0 0 r 0.00 00 % 0 0 0 0 0 o ; ,o 0 0 000 l0 m 0 5 lmIn 1 I I I 20 I ,8 0 5 10o 15 3 U ,I 14 n , /; on October 1, 2021 by guest. Protected copyright. I 1. ai~~~~~ 9 z 1000 m IS0 m Om IxVI nr 84PArl in 15 0 5 1010 15 0 5 tO Fig. 1: Clustering ot runners at the 9.8 km point in the Fukuoka Marathon, as estimated from passing times on a video record. The distribution of competitors within eight clusters is illustrated. RESULTS The density of the clusters increased with running speed, Address for correspondence: from a value of 0.44 runners.m2 at a pace of 4.8 m.s, to 0.80 Professor Roy J. Shephard, Director runners.m2 at 5.6 m.s (Fig. 2). For 95% of the clustered School of Physical & Health Education runners, the shortest distance was in the 320 Huron Street interpersonal Toronto, range 0.5-1.5 m; frequency distribution curves for distances Ontario M5S lAl to the first, second, and third nearest rivals are shown in Canada Fig. 3. 167 1.2 *3---0 Tokyo Marathon * Fukuoka Marathon Br J Sports Med: first published as 10.1136/bjsm.21.4.166 on 1 December 1987. Downloaded from A Marathon Fukuoka o Tokyo Marathon 0 1.0 i '0315X - 1 030 30m .0'793 _. SM0 X- 2052 'IE A .'0. 7a8 -15 20.1l5 *20 / i0.6 I /£LA 0 ,P9:14 \ ~~ Io 0 _: . \ /f 0 0 / 64*0*0 '00 O ° oe *fr I \ * %0~\\ .,0X~4 20 2-r 0% 4-0 50 60 0~~~~~ Running Speed (rmnsec) Fig. 2: The relationship between the density of competitor clusters and .O . running speed. Data for the Tokyo and Fukuoka marathon races, with least N o_ - squares regressions. o~~~~o t 40 - fr-W1- st paufl dostanc - X d shWtet Pamu dstance 30 Fig. 4: The shortest inter-personal distance and the corresponding angular separation of the nearest rival relative to the direction of running. Points ~20 represent location of individual relative to direction of race (shown by arrow) and location of rival (assumed at intersection of horizontal and vertical axes). 10~~~~~~~. x running and (4) mutual supervision and surveillance over 10 0 the course of the event. Against these potential benefits must be weighed possible disadvantages, including (1) a x ~~~~~~0.x disturbance of personal running x rhythms, (2) conformity http://bjsm.bmj.com/ 0 o---o with what may be an unsuitable running pace for the individual and (3) a possible 04 0.8 1.2 2.0 24 2-8 decrease of mechanical t6 efficiency due to (1) and (2). The Shortest Personal Distance (m) It is particularly interesting that clustering is most marked among the elite performers. This Fig. 3: Frequency distribution curves ofinter-personal distances separating a may reflect both competitor from the first, second and third closest rivals. the greater importance of reducing wind resistance at higher speeds (resistance varies as the cube of velocity, on October 1, 2021 by guest. Protected copyright. The angle separating a competitor from his nearest rivals Margaria, 1976; Pugh, 1976; Shephard, 1982) and the is illustrated in Fig. 4; 95% of other competitors avoided greater competitive stimulus in a fast moving group. occupying the angle ± 150 ahead and behind a given Factors (3) and (4), on the other hand, probably assume runner. This was probably partly to avoid collision if the greater importance for those participants who are at a lower front runner slowed his pace and partly because it was level of competition. difficult to see the track when another person was directly The phenomenon of clustering among long-distance ahead. runners does not appear to have been noted previously but merits further study by those interested in marathon DISCUSSION performance. Our data support the view that marathon runners separate themselves from their nearest rivals by a relatively constant distance and angle throughout a race and further show that a clustering of competitors occurs at a variety of running speeds. References The anticipated tactical advantages from the observed Margaria, R., 1976. Biomechanics and Energetics of Muscular Exercise. grouping behaviour include (1) a reduction of air resistance, Clarendon Press, Oxford. (2) an upsurge and maintenance of competitive spirit, (3) Pugh, L. G. C. E., 1976 "Air Resistance in Sport" pp. 149-164. In: Advances in Exercise Physiology. Eds.: E. JokI, R. L. Anand and H. Stoboy. Karger, resolution of uncertainties regarding an appropriate Basal. running pace at various points in the race, with a Shephard, R. J., 1982. Physiology and Biochemistry of Exercise. Praeger, consequent maximisation of mechanical efficiency of New York..