© 2005 Nature Publishing Group

NATURE|Vol 437|27 October 2005 BRIEF COMMUNICATIONS ARISING

E10

The shirt colour worn by sportsmen can affectthe behaviour of the competitors1,2, but Hilland Barton3 show that it may also influencethe outcome of contests. By analysing theresults of men’s combat sports from the Athens2004 Olympics, they found that more matcheswere won by fighters wearing red outfits thanby those wearing blue; they suggest that redmight confer success because it is a sign ofdominance in many animal species and couldsignal aggression in human contests. Here weuse another data set from the 2004 Olympicsto show that similar winning biases occur incontests in which neither contestant wears red,indicating that a different mechanism may beresponsible for these effects.

If, as Hill and Barton claim, there is some-thing special about the colour red, then con-tests using other colour pairings should not bebiased. We tested whether this could be thecase by analysing data (www.athens2004.com)from contests in which red was not used forcompetitors’ outfits: in judo matches, oneplayer wears blue and the other wears white.We followed the methodology originally usedby Hill and Barton3 for boxing, tae kwon do,and Greco–Roman and freestyle wrestling.

After ensuring that outfit (judogi) colour in judo contests was allocated at random(www.ijf.org), we found a significant winningbias for players wearing blue compared withthose wearing white (�2�7.34, d.f.�1,P�0.01), and a similar effect of contest sym-metry on winning bias to that reported by Hilland Barton (Fig. 1). We also found the win-ning bias for players wearing blue when con-sidering only contests in the first round ofcompetition (�2�4.85, d.f.�1, P�0.05). Thisresult excludes the possibility that theobserved bias might arise through skilled con-testants being placed, by chance, in draw posi-tions where they wear blue more often as theyprogress through the competition.

Our results indicate that there is nothing

inherently special about red in terms ofcolour-associated winning biases.

We can think of no plausible evolutionaryexplanation based on animal behaviour orevolutionary psychology that might accountfor a winning bias for blue contestants. Wepropose instead that outfit colour affectsopponent visibility, which is crucial for avoid-ance and interception, and for anticipatingbehaviour. Visual abilities that could influencesporting performance include being able to

follow rapidly moving objects and performfast visual searches4–6. And the hue, saturation,brightness and contrast of an object (or oppo-nent) could enable it to be picked out againstits background7,8. These factors are critical forcombat sports and for detecting teammates onthe field of play (http://www.liv.ac.uk/researchintelligence/issue1/manunit.html; 1999).

In judo, the white judogi is likely to be per-ceived as brighter than the blue and may havehigher contrast against the background. Menwearing blue may therefore have a visualadvantage in being able to anticipate their(white) opponents’ moves. We do not knowthe reflectance spectra, lighting arrangementsor other visual factors that might have affectedthe visual salience of the red and blue outfitsworn in the sports studied by Hill and Barton.Although our hypothesis is untested, visibilitydifferences could also explain the biases theyfound. The visual attributes of sporting wearshould therefore be considered in this widerperceptual context.Candy Rowe*, Julie M. Harris†, S. Craig Roberts‡*School of Biology & Psychology, University ofNewcastle, Newcastle upon Tyne NE2 4HH, UKe-mail: candy.rowe@ncl.ac.uk†School of Psychology, University of St Andrews,St Mary’s College, St Andrews KY16 9JP, UK ‡Evolutionary Psychology & Behavioural EcologyGroup, School of Biological Sciences, University ofLiverpool, Liverpool L69 7ZB, UK

1. Frank, M. G. & Gilovich, T. J. Pers. Soc. Psychol. 54, 74–85(1988).

2. Mills, B. D. & French, L. M. J. Hum. Mov. Stud. 31, 47–60(1996).

3. Hill, R. A. & Barton, R. A. Nature 435, 293 (2005).4. Williams, A. M. J. Sport Sci. 18, 737–750 (2000).5. Farrow, D. & Southgate D. Clin. Exp. Optom. 83, 226–231

(2000).6. Williams, M. Psychologist 15, 416–417 (2002).7. McKeefry, D. J. et al. Invest. Ophthal. Vis. Sci. 44, 2267–2276

(2003).8. Regan, B. C. et al. Phil. Trans. R. Soc. Lond. B 356, 229–283

(2001).

doi:10.1038/nature04306

SPORTING CONTESTS

Seeing red? Putting sportswear in contextArising from: R. A. Hill & R. A. Barton Nature 435, 293 (2005)

0.4

0.5

0.6

Degree of asymmetry

0.3

0.7

Prop

ortio

n of

con

test

s w

on

All None Small Medium Large

Figure 1 | Influence of judogi colour on theoutcome of judo matches in the Athens 2004Olympics. The black line at 0.5 indicates theexpected proportion of wins by blue or whiteunder the null hypothesis that colour has noeffect on contest outcome. Blue bars, proportionwon by players in blue; white bars, proportionwon by players in white. There were significantdifferences between the number of blue andwhite wins for all contests combined (� 2�7.34,d.f.�1, P�0.007). This difference is mostevident at low degrees of asymmetry in relativeability of the two competitors in each bout(subdivided using quartiles of points difference,after Hill and Barton3). Differences between thenumber of white and blue wins were as follows:no asymmetry (� 2�1.53, d.f.�1, P�0.22), smallasymmetry (� 2�6.31, d.f.�1, P�0.012),medium asymmetry (� 2�3.19, d.f.�1, P�0.07)and large asymmetry (� 2�0.57, d.f.�1,P�0.81). Number of contests recorded are 301,79, 99, 53 and 70, respectively.

SPORTING CONTESTS

Hill & Barton replyReplying to: C. Rowe, J. M. Harris and S. C. Roberts Nature 437, doi:10.1038/nature04306 (2005)

Rowe et al. corroborate our finding that thecolour of clothing influences the outcome ofsporting contests1, but they offer a differentmechanism to explain the effect. We foundthat in four combat sports wearing red wasconsistently associated with improved perfor-

mance relative to wearing blue2, and arguedthat wearing red enhances performancethrough psychological effects on the wearerand/or on the opponent. We suggested thatthese psychological effects reflect the evolu-tionary and cultural associations of red with

dominance and aggression. Rowe et al. findthat, in a fifth combat sport, wearers of blueoutperformed wearers of white. They attributetheir and our results to perceptual rather thanto psychological effects, arguing that visibilityof the opponent is the critical factor.

In our view, this visibility explanation isunlikely in a situation where contestants fightat close quarters in brightly lit arenas, as inthese combat sports. Crucially, in the combatsports we analysed, the hypothesis of Rowe et al. requires that blue-wearing opponents bemore visible than their red-wearing opponents,

© 2005 Nature Publishing Group

BRIEF COMMUNICATIONS ARISING NATURE|Vol 437|27 October 2005

E11

but, insofar as photographs provide an indica-tion of this, the opposite seems to be true (seehttp://www.athens2004.com/en/BoxingImageGallery/imagegallery). Furthermore, the liter-ature on visibility effects cited by Rowe et al.refers to sports where visuospatial judgementsare made over distance. We also note that visi-bility has not been used to explain similareffects of artificially coloured leg bands ondominance interactions in birds3.

The visibility explanation proposed byRowe et al. suggests that male and female con-tests should be equally affected by colour,whereas our sexual-selection hypothesiswould predict stronger effects for males. Thelatter prediction is supported by the data: inthe same sports in which significant effectswere found for males, no such effects are evi-dent for females (tae kwon do and freestylewrestling (red against blue): �2�0.32, n�155,d.f.�1, P�0.50; judo (blue against white):�2�0.30, n�214, d.f.�1, P�0.50). Pooling

the data for the three combat sports in whichboth men and women participated, there is asignificant association between sex and pro-portion of contests won by the advantageouscolour (�2�6.44, d.f.�1, P�0.011).

We suggest that the impact of colour oper-ates through its psychological and hormonalinfluences, rather than through its effects onvisibility. Why then should blue provide anadvantage over white? Although red reflectscurrent dominance and testosterone levels4

and indicates emotional arousal5, other brightcolours, including blue, are known to indicatelong-term developmental vigour6 and maytherefore have psychological effects relative to paler colours or white. In any case, ourdemonstration of enhanced performance inred relative to blue and other colours across arange of conditions suggests that, contrary toRowe et al.’s claim, there is indeed somethingspecial about red.

Identification of the precise mechanisms

underlying the effects of colour will requiremore detailed experimental work. Mean-while, the results of Rowe et al. lend furthersupport to our claim that colour needs to betaken into account in ensuring a level playingfield in sport.Robert A. Barton, Russell A. HillEvolutionary Anthropology Research Group,Department of Anthropology, Durham University,Durham DH1 3HN, UKe-mail: r.a.hill@durham.ac.uk

1. Rowe, C., Harris, J. M. & Roberts, S. C. Nature 437,doi:10.1038/nature04306 (2005).

2. Hill, R. A. & Barton, R. A. Nature 435, 293 (2005).3. Cuthill, I. C. et al. Proc. R. Soc. Lond. B 264, 1093–1099

(1997).4. Setchell, J. M. & Wickings, E. J. Ethology 111, 25–50

(2005).5. Drummond, P. D. & Quah, S. H. Psychophysiology 38,

190–196 (2001).6. Andersson, M. Sexual Selection (Princeton Univ. Press,

Princeton, 1994).

doi:10.1038/nature04307