One clear trend for all time-based athletic events, including running, cycling, swimming, rowing, speed skating, and cross-country skiing, is that performance for all competition distances has steadily improved during the last 75 yr. A second trend is that the rate of improvement for females has exceeded that of males during most of that time period. For example, in 1936, Jesse Owens ran 100 m in a world-record 10.2 s. The same year, Helen Stephens lowered the women's record to 11.5 s. In 2006, the men's 100-m record stood at 9.77 s, and the women's record was 10.49 s. This represents more than double (4.2 vs 8.8%) the rate of improvement in the men's and women's records, respectively. Consequently, despite firmly established sexual dimorphisms in skeletal muscle mass and strength (12,14,16,25), anaerobic power and capacity (18,19), and maximal aerobic capacity (1,2,10,22), some scientists have analyzed the trend of world records or winning performances in running over time and have used linear regression lines to predict that the gender gap in physical performance will eventually disappear and that the best females will someday compete equally with the best males (11,23,29). Others have analyzed the historical data for running and have concluded that the gender difference is no longer narrowing (6,15). Outside of the obvious interest in historical trends, such analyses raise the interesting question of the biological versus cultural determinants of sexual dimorphism and how advances in training methodology, equipment technology, cultural acceptance for women in sport, and various pharmacological agents have contributed to performance developments during the last half century.
Given the equivocal interpretations of historical and current performance trends, the present study was performed to quantify the gender difference in human performance in sports where anaerobic muscular power output is of particular importance. We analyzed Olympic and World Championship performances from the last five decades in running, swimming, and speed skating for distances requiring 10-60 s for completion. Our analytical approach has allowed us to specifically address the question: Does the historical performance record indicate that the gender difference in human anaerobic performance is still narrowing, is now stable, or is now wider than previously observed?
For analysis, performances were chosen for three sports in which both males and females have competed in international championship finals since at least the 1950s or early 1960s, at the same venue and approximately same date: sprint athletics (100 m, 200 m, 400 m,), speed skating (500 m), and swimming (100-m freestyle and 100-m backstroke). These events have approximate competition times of 10-60 s, making peak muscular power and anaerobic power and capacity major performance determinants. In addition, they are contested such that all participants attempt to achieve the fastest time possible, as opposed to a tactical "race to win and not against the clock" strategy that makes championship finals in the longer events less representative of current best-performance standards. Official results were collected and verified from publicly available sources such as governing-body Web pages (e.g., www.iaaf.org), personal communications with sports historians and archivists, and published compilations (26). Instead of basing the analysis on the times of winners alone, the top six finalists at each world championship or Olympic final were included. In both running and swimming, finals events have eight competitors. However, it is not uncommon for disqualifications to occur because of false starts or injury. Therefore, only the first six finishers were included for analysis. The year 1952 was chosen as the starting time point for all analyses. However, women's speed skating Olympic finals were not added to the Olympic program until 1960, and the women's 400-m running event was not added until 1964. We argue that the last five decades of competition can be viewed as the modern era of international sport. Although not totally free of world conflicts, this period has been characterized by strong national team organization, rigorous selection procedures, and consistent participation of each nation's best athletes.
Official results for all championship finals in the 100-, 200-, and 400-m athletics events held between 1952 and 2005 were collected from public sources (11-14 Olympic plus 10 world championships: 1983, 1987, 1991, 1993, 1995, 1997, 1999, 2001, 2003, and 2005). Times determined from archival film analysis of early hand-timed 100- and 200-m events were used to ensure comparable (electronic) performance times during the period (26). To ensure that the gender difference in performance was as accurate as possible, 100- and 200-m results were also corrected for wind and altitude using published equations (27,28). The time correction applied for 200 m was based on the wind velocity and direction in the 100-m straight section of the race. In two finals during this period, external confounders were identified that so clearly impacted the final results that they were excluded from the overall analysis (1952: significant wind present but not measured during the men's 100-m Olympic final; and 1960: rainstorm during the women's 200-m Olympic final). The political boycotts of the 1980 (United States, West Germany, Japan, and other Western countries) and 1984 (USSR, GDR, and other Eastern Bloc countries) Olympics were also considered potential confounders. However, because the impact of the boycotts on the participation of the best athletes varied from event to event, we ultimately chose to include these finals for all distances.
100-m freestyle and backstroke swimming.
Official results for all 26 long-course (50 m) championship events held between 1952 and 2005 were compiled from public sources and historical archives. In addition to Olympic finals, long-course world championships have been held since 1973 (1973, 1975, 1978, 1982, 1986, 1991, 1994, 1998, 2001, 2003, and 2005) and were also included in the analysis. The 100-m distance was chosen for analysis instead of the 50-m event because the shorter freestyle distance was not added to the Olympic program until 1988. The freestyle and backstroke events were the only short-distance events in which both men and women have competed since the 1950s.
500-m speed skating.
Although women competed in official all-around championships (combined results from 500-, 1500-, 3000-, and 5000-m events) in speed skating beginning in 1936, women's speed skating was not added to the winter Olympic program until 1960. Single-distance world championships were not introduced until 1996. Therefore, 21 events, 13 Olympic finals results since 1960, and eight annual single-distance world championship results since 1996 were collected from public archives and were combined for analysis. The 1000-m distance was not included in the analysis (despite an event duration just over 60 s) because men did not compete in this event until 1976.
Conversion of performance times to average power.
On the basis of power-balance models for running, speed skating, and swimming, as presented by de Koning and van Ingen Schenau (7), Arsac and Locatelli (3), de Koning et al. (8), and Toussaint et al. (24), the average power of each studied athletic performance was estimated. It was assumed that the drag parameters in the models were constant during the studied time interval. In the case of the 100-m backstroke power estimation, drag parameters derived from published models for the freestyle stroke (24) were applied, in the absence of published drag estimations for the backstroke technique. Because the chosen parameter values modeled contemporary athletes, some underestimation of average power can be expected in the early decades of our analysis. This underestimation will impact only the absolute average power-output values, not the relative gender difference in power.
The data analyzed comprise a population of international finalists. We have, therefore, not applied inferential statistics to compare performances in males and females. After corrections and deletions were performed as described, performance results for men and women for each sport were analyzed. Regression lines were applied to the historical time series using Microsoft Excel. A linear regression line (y = ax + b) was fitted to each time series, followed by a second-order polynomial regression (y = ax2 + bx + c). An a priori decision was made that where the R2 for goodness of fit improved by 0.02 or more with second-order polynomial regression, this curve was retained as the line of best fit for the time series. Otherwise, a linear regression model was retained as the line of best fit for the time series. For each placement position (first to sixth) in each male and female final pair, the percentage difference in performance time between female and male performances was calculated using the following formula:
Performance differences for each championship final were combined in a separate time series for each event and were also analyzed as described above. Finally, parallel analyses based on the estimated average power output for each performance were performed using the same approach.
In the 100-m sprint, male performances were best fit by a linear regression, with males improving approximately 0.01 s·yr−1 for five decades (Fig. 1, panel A). Hundred-meter finals performances over time for females were best fit by a second-order polynomial, because absolute performances in the 100-m final for women have marginally deteriorated during the last 15 yr. The gender difference in 100-m sprint final performances decreased from 12.9% in the 1950s to a low of 8.6% in the 1980s (Fig. 1, panel B). In the period 2000-2005, the gender difference averaged 10.3%. Analysis of the 200-m race revealed a similar trend. However, here, both male and female performances over time were best fit by a second-order polynomial regression line (Fig. 1, panel C). The behavior of the gender difference in 200-m performance (Fig. 1, panel D) was essentially identical to that observed for the 100-m final, decreasing from 14.5% in the 1950s to a nadir of 9.6% in the 1980s and then increasing to a current value of 11.4% for 2000-2005. Male and female performances in the 400-m races were best fit by a second-order polynomial. For male performances, the improvement in fit compared with a linear regression was marginal (r2 0.26 vs 0.22). The slopes of the regression lines demonstrate that performance in this event has been comparatively stable during the last 30 yr (Fig. 1, panel E). The gender difference in performance was 17.4% in the 1960s, averaged 11.2% in the 1980s, and rose to 11.9% for 2000-2005 (Fig. 1, panel F). For all three running events, the performance difference between males and females averaged 9.8% between 1980 and 1988, and 11.2% between 2000 and 2005.
Results for 100-m freestyle and backstroke swimming finals are presented in Figure 1 (panels G-J). Both men's and women's results were well fit by a second-order polynomial, revealing a marked decline in the rate of improvement in performance over time. The gender difference in performance time for the 100-m freestyle was also best fit by a second-order polynomial and decreased from 14.3% in the 1950s to a low of 11.5% in the 1980s. Since 2000, the gender gap in sprint freestyle swimming performance has increased to 12.2% (Fig. 1, panel H). The 100-m backstroke data showed the same trend. Both male and female performances were well fit by a second-order polynomial regression line (Fig. 1, panel I), with the gender difference decreasing from 13.9% in the 1959s to 10.8% in the 1980s, before increasing to 12.04% for 2000-2005 (Fig. 1, panel J). For the two swimming events combined, the gender difference averaged 14.1% in 1952-1956, 11.2% in 1980 to 1988, and 12.1% from 2000 to 2005.
Results for the 500-m speed skating final are presented in Figure 1, panel K. In speed skating, both male and female performance trends were well fit by a linear regression line, demonstrating that speed skating performances are still improving rapidly compared with running and swimming. And, overall, the slope of the female regression line was steeper compared with that of males. However, when performance differences were calculated, their pattern over time was best fit by a second-order polynomial (Fig. 1, panel L), suggesting that the decline in performance difference between male and female speed skaters has also reversed during the last decade.
This study provides evidence from three different sports that the gender difference in elite human performance is no longer narrowing for events lasting approximately 60 s or less, a duration range for which muscular power and anaerobic energy metabolism play a major role. Our results are in sharp contrast to those of several previous studies that have predicted an eventual eradication of sexual dimorphism as a variable in elite sprint performance (11,23,29). Our results are, however, consistent with the underlying physiological differences expressed by the male and female genotypes. More surprisingly, our data also demonstrate that the gender gap in elite sprint performance has actually widened from its nadir observed in the late 1970s and to the end of the 1980s. This same pattern emerges within the overall performance development timeline of the most explosive, anaerobic events in running, swimming, and speed skating, despite clear differences in both international participation and the potential contribution of technological innovation to performance changes over time. For all three sports, the trend of a widening gender difference seems to be attributable to a greater relative change in performance development among females. The rate of improvement has slowed more for females than males for swimming and speed skating. In running 100-400 m, average performance has actually deteriorated in females at the highest levels of competition compared with the best performances of the 1970s and 1980s.
Power output in a wide variety of athletic events is well known to be related to an exponential expression of velocity. The magnitude of difference in power output associated with a given performance-time difference gives a more correct representation of underlying differences in physiological capacity. We found differences in power output in the range of 20-30% for running and speed skating and approximately 45% for swimming (Fig. 2), which is consistent with reported gender differences in lower- and upper-body muscle mass and maximal strength (12,14,16,25). Calculations of the power requirement of the observed performances using existing models for running, swimming, and skating (3,7,8,24) show similar results with respect to relative gender differences as does performance time-that is, a widening gender gap largely attributable to a relative decrease in calculated power output by female performers. Assuming that technological innovations have the same impact on male and female performance, the observations of a widening gender gap during the last decade or more suggest that the nadir of gender differences in performance observed in the 1970s and 1980s was, to some degree, artificial. These observations also are consistent with the interpretation that what is suspected to have been widespread and, in some cases, systematic, doping (5,13) contributed to the reduction in gender differences observed from 1952 through the late 1980s.
Presently, it remains unclear what the natural gender difference in human anaerobic performance should be. In the approximately 75-yr history of comparable men's and women's performances on the world stage, we have probably not witnessed a time when female athletes were equally encouraged to compete in sport, were trained as well as men, and were not doping. Given the nature of the results, it is reasonable to first focus on the historical curve for female performances and ask why the relative reversal (and absolute reversal for running) has occurred in females. There is no evidence to suggest that a decline in female participation at the highest levels in sports can account for the performance trends observed. For example, women's participation in the summer Olympics has increased steadily from 518 females and 11% of total participants in 1952, to 2186 and 26% in 1988, to more than 4000 female competitors and approximately 40% of total participants in 2004. Because the number of competitors per event from a given country is restricted, this increase reflects female athletes from more countries competing in more events. The Olympic Games and world championships have become more representative of world's best over time. Because all of the competitions were held at the same venues, systematic differences in conditions over time, such as faster track surfaces, wave-damping pools, or better ice quality on indoor tracks, would be expected to impact both males and females equally. Random variations in the conditions, such as wind, have been partially accounted for in the running analysis, but not in the swimming or speed skating analyses. However, during the last 20 yr, the majority of both swimming and speed skating international championships have been held indoors under well-controlled conditions. Therefore, a run of random fluctuations in the last 15 yr that might have selectively impacted female performances more than males in three different sports is an extremely unlikely explanation for the findings presented. Finally, in modern international athletics, swimming, and speed skating, males and females often train together and have equal exposure to all training and technique developments. The integration of men's and women's training at the national team level has only increased in recent years, making a widespread deterioration of the training environment for elite female athletes an unlikely explanation for the observed performance changes.
The change in the performance environment of international sport that best coincides chronologically with the performance trends observed here is the intensification of the scope and sensitivity of drug testing among elite athletes. More athletes are being tested, both out of competition and in competition, for a broader spectrum of illegal agents, with greater sensitivity. As early as 1928, the International Amateur Athletic Federation became the first international sport federation to ban performance-enhancing drug use. However, with no testing in place, the ban was unenforceable. A test for anabolic steroids was not developed until 1974, and steroids were not placed on a banned-substance list until 1976. Meanwhile, the 1970s and 1980s were marked by rumors of systematic state-sponsored drug use by elite athletes, particularly among those from former Eastern Bloc nations. This type of abuse has subsequently been documented extensively for the former German Democratic Republic, with the drug regimens of numerous champion female athletes in both swimming and athletics provided in archival government documents (13). Though not as robust, there is also documentation of similar practices in the former Soviet Union (17). And Eastern Bloc countries were not the only international sports powers whose athletes used performance-enhancing drugs (5,21). One physician from California testified before a U.S. Senate committee that he had prescribed anabolic steroids to 20 medal winners of the 1984 Olympic Games (5). Indeed, the most internationally visible positive drug test was on Canadian athlete Ben Johnson, who won the gold medal in the 100-m sprint for Canada at Seoul, Korea in 1988. This disqualification accelerated antidoping efforts. At the same time, the collapse of the communist governments of sports powers East Germany, the Soviet Union, and others ensued, which presumably impacted the training (and doping) programs of some of the best athletes from those countries. Antidoping efforts were further intensified with the formation of the World Anti-Doping Agency (WADA) in 1999, in the aftermath of drug scandals in the Tour de France in 1998. Today, WADA performs unannounced, out-of-competition tests (OOCT) on all international-level athletes. For example, in 2004, 2327 OOCT were performed on international-caliber athletes in 62 countries. A total of 19 adverse findings and four refusals were reported (WADA Annual Report 2004 (http://www.wada-ama.org/rtecontent/document/WADA_2004_Annual_Report_en.pdf)). Of these tests, a total of 558 drug tests were performed on athletes in aquatics, athletics, and ice skating (2004 WADA OOCT Testing Statistics (http://www.wada-ama.org/rtecontent/document/2004_stats.pdf)). Thus, the chances of a top-10 international performer not being subjected to out-of-competition testing during a given year are now quite small. Some athletes are, indeed, tested multiple times in a season. These tests are performed in addition to in-competition testing and the testing that national governing bodies perform. At least one published study has concluded that an increase in the test frequency of doping tests was associated with a gradual decline in the percentage of positive samples in targeted sports (4). Although performance-enhancing drug use almost certainly still occurs, it is clear that athletes contemplating doping face far more rigorous and sensitive controls today than they did 20 yr ago, at the time of the nadir of the relative performance difference between genders.
If we accept that drug-testing improvements have played a role in the relative decline in performance seen among elite females, it is natural to question why a parallel effect is not readily evident for male performances. We speculate that the improvement in training adaptations among females using testosterone or testosterone surrogates as doping agents is greater than the advantage to be gained by men, who already have high levels of naturally occurring testosterone. An alternative explanation would be that doping testing has proven more effective in reducing drug use in female performers than males because the endogenous hormone profile for females represents a lower baseline and, thereby, a less ambiguous chemical background for comparison. However, we have no specific evidence to support this explanation. The gender breakdown of adverse findings in WADA out of competition testing is not reported. Even if one were available, a gender breakdown for positive drug tests would not provide information about gender differences in the prevalence of false-negative reports.
The present analyses also demonstrate that at the highest levels of running and swimming, performance development is stagnating. In that regard, our interpretation coincides with the recent findings of Nevill and Whyte (20), who found that the world-record progression in several running events was well described using a sigmoidal regression curve. We chose to analyze finalist performances in championship events instead of singular world records. The advantages of this approach are that 1) it gives a larger number of data points from which to deduce trends over time, and 2) it provides the possibility of detecting a performance decline that would not be detectable from world records alone. We have analyzed performance developments in three different sports to determine whether the trends observed in sprint running were unique. The present findings, and those of other recent studies (6,20), suggest that the effects of increased recruitment, training advances, and technological developments may be reaching a saturation point. The three sports analyzed here differ with respect to 1) international participation and 2) potential impact of technological innovation. Sprint running is characterized by broad international participation combined with relatively little room for technological or materials science innovation. We would expect that saturation effects for performance-enhancing developments would first become evident in running, as appears to be the case. It is noteworthy that whereas 100-m sprint performance by males was well fit by a linear equation with a negative slope, performances during the last 15 yr are very stable. In contrast, speed skating has a relatively low international participation base. At the same time, the impact of advances such as indoor skating ovals, reductions in ice friction, aerodynamically improved suits, hinging of the skate blade (9), and carbon fiber-molded skating boots all have contributed to velocity improvements. Consequently, speed skating can be viewed as relatively immature, with substantial potential for further development. Swimming shares characteristics of both these extremes, with its broad international participation coupled with substantial technological impact. Thus, although we recognize that history has previously proven scientists to be foolish in predicting that the ultimate in human performance has been achieved, the pattern of performance data in sports such as sprint running suggests that we may be approaching the limits possible within the current genome.
Finally, it is important to recognize that performance power increases exponentially with performance velocity for all three sports. Therefore, in the absence of radical reductions in power losses or mechanical efficiency attributable to technological innovation, advances in performance time attributable to enhancements in human power capacity would be expected to show clear plateau effects as the benefits of secular increases in adult body size and tolerable training loads are maximized. This point seems to be at hand for the sports of running and swimming.
The modern era of sport has been marked by increased international participation, advances in athlete preparation, and steady improvements in performance. The more rapid rate of improvement observed among elite females compared with that of males has proved to be enticing fuel for speculation among some sport scientists and other observers. In the last 20 yr, several scientists have used performance developments to defend a prediction of an eventual eradication of physiologically based gender differences in performance. The results of the present study clearly refute those predictions. More surprisingly, these data also show clear evidence of a recently widening gender gap in performance in short-duration, high-intensity events. This finding is both reasonably explained by and provides indirect evidence for the recent success of improved international doping testing in deterring athletes from using illegal performance-enhancing agents.
1. Armstrong, N., B. J. Kirby, A. M. McManus, and J. R. Welsman. Aerobic fitness of prepubescent children. Ann. Hum. Biol.
2. Armstrong, N., J. R. Welsman, and B. J. Kirby. Peak oxygen uptake and maturation in 12-yr olds. Med. Sci. Sports Exerc.
3. Arsac, L. M., and E. Locatelli. Modelling the energetics of 100 m running by using the speed curves of world champions. J.Appl. Physiol.
4. Bahr, R., and M. Tjornhom. Prevalence of doping in sports: doping control in Norway, 1977-1995. Clin. J. Sports Med.
5. BIDEN, J. R. Steroids in Amateur and Professional Sports: The Medical and Social Costs of Steroid Abuse. U.S. Senate Committee on the Judiciary Hearing J-101-102, Washington, DC, 1990.
6. Cheuvront, S. N., R. Carter III, K. C. Deruisseau, and R. J. Moffatt. Running performance differences between men and women. Sports Med.
7. De Koning, J. J., and G. J. Van Ingen Schenau. On the estimation of mechanical power in endurance sports. Sport Sci. Rev.
8. de Koning, J. J., C. Foster, J. Lampen, F. Hettinga, and M. F. Bobbert. Experimental evaluation of the power balance model of speed skating. J. Appl. Physiol.
9. de Koning, J. J., H. Houdijk, G. de Groot, and M. F. Bobbert. From biomechanical theory to application in top sports: the klapskate story. J. Biomech.
10. Dill, D. B., L. G. Myhre, S. M. Greer, J. C. Richardson, and K. J. Singleton. Body composition and aerobic capacity of youth of both sexes. Med. Sci. Sports
11. Dyer, K. F. The trend of the male-female differential in various speed sports 1936-84. J. Biosoc. Sci.
12. Ford, L. E., A. J. Detterline, K. K. Ho, and W. Cao. Gender- and height-related limits of muscle strength in world weightlifting champions. J. Appl. Physiol.
13. Franke, W. W., and B. Berendonk. Hormonal doping and androgenization of athletes: a secret program of the German Democratic Republic government. Clin. Chem.
14. Fuster, V., A. Jerez, and A. Ortega. Anthropometry and strength relationship: male-female differences. Anthropol. Anz.
15. Holden, C. An everlasting gender gap? Science
16. Janssen, I., S. B. Heymsfield, Z. M. Wang, and Z. Ross. Skeletal muscle mass and distribution in 468 men and women aged 18-88 yr. J. Appl. Physiol.
17. Kalinski, M. I. State-sponsored research on creatine supplements and blood doping in elite Soviet sport. Perspect. Biol. Med.
18. Maud, P. J., and B. B. Shultz. Gender comparisons in anaerobic power and anaerobic capacity tests. Br. J. Sports Med.
19. Murphy, M. M., J. F. Patton, and F. A. Frederick. Comparative anaerobic power of men and women. Aviat. Space Environ. Med.
20. Nevill, A. M., and G. Whyte. Are there limits to running world records? Med. Sci. Sports Exerc.
21. Scarpino, V., A. Arrigo, G. Benzi, et al. Evaluation of prevalence of "doping" among Italian athletes. Lancet
22. Sharp, M. A., J. F. Patton, J. Knapik, et al. Comparison of the physical fitness of men and women entering the U.S. Army: 1978-1998. Med. Sci. Sports Exerc.
23. Tatem, A. J., C. A. Guerra, P. M. Atkinson, and S. I. Hay. Athletics: momentous sprint at the 2156 Olympics? Nature
24. Toussaint, H. M., G. de Groot, H. H. C. M. Savelberg, K. Vervoorn, A. P. Hollander, and G. J. van Ingen Schenau. Active drag related to velocity in male and female swimmers. J.Biomech.
25. Vanderburgh, P. M., M. Kasuno, M. Sharp, and B. Nindl. Gender differences in muscular strength: an allometric model approach. Biomed. Sci. Instrum.
26. Wallechinsky, D. The Complete Book of the Summer Olympics
. Athens 2004 ed. Wilmington, DE, 2004: Sport Media Publishing, pp. 204-219.
27. Ward-Smith, A. J. Air resistance and its influence on the biomechanics and energetics of sprinting at sea level and at altitude. J. Biomech.
28. Ward-Smith, A. J. A mathematical analysis of the influence of adverse and favourable winds on sprinting. J. Biomech.
29. Whipp, B. J., and S. A. Ward. Will women soon outrun men? Nature