It is no longer a matter of if but rather when.

BY HOLLY ORTLUND, MS; WILL HUDSON, BS; and DAVID MARTIN, PhD

anew world record (2:03:38 by Patrick Makau, September 25, Berlin) every one of the World Marathon Major course records was broken (London, Boston, Berlin, Chicago, and New York). Such impressive finishing times over the past 12 months have focused the world’s attention on the marathon much as the mile entranced the world of sport more than 60 years ago. With seven performances under the 2:05 mark in 2011, the once-deemed-impossible barrier of a human covering the 26.2-mile (42.195K) distance in less than two hours now seems not only feasible but expected. Achieving this feat requires running a seemingly inconceivable pace of 4:35 per mile (2:51 per kilometer), or viewed more practically to some, running back-to-back 68-second splits for 105 1/2 consecutive laps around a 400-meter track. For comparison’s sake, Geoffrey Mutai’s world-leading effort in Boston last spring equates to an average of 4:42 per mile (2:55 per kilometer) and 70-second lap splits.

Humans seem particularly drawn to clean, round numbers, and the fact that we are now less than four minutes away from someone surpassing the two-hour barrier has led, not surprisingly, to a great deal of recent attention on the marathon distance. Achieving such an act is a topic of curiosity for many, and thus the intrigue of predicting or projecting future running records is not new. Numerous articles have been published on the subject. Various statistical models have been developed, but the best approach has been to analyze running performance over time with asymptotic functions, although linear and polynomial models are also used. With enough data points, these “time versus year” plots yield curves that can be used to predict world records as well as ultimate physiological performances. Liu and Schutz (1998) analyzed marathon world records and best performances

ye year 2011 was an incredible one for marathon performances. Along with

per year across several running distances and models (linear and nonlinear) and it seemed that an exponential-decay model relating running time and historical year was the most valid deterministic model for predicting performances.

To predict the date of the first sub-two-hour marathon, we compiled the fastest annual performance (FAP) times from 1924 to 2011 as well as the world-record progression (WR) during this same period (see tables 1, 2, and 3 on pages 29 to 35 of this article). A one-phase exponential decay model was used to fit the data, enabling us to predict both the year when the two-hour barrier will be broken as well as the fastest theoretical marathon time possible.

We analyzed three different marathon data sets. Two (tables 2 and 3) are slightly varying lists of the fastest annual marathon times (the World War II years of 194345 were removed from statistical analysis) and include data from David Martin, former statistician for the Association of International Marathons and Distance Races, or AIMS, as well as data from the Association of Road Racing Statistics, or ARRS. We chose 1924 as the starting point because this was the year the official [AAF marathon distance of 42.195K was initially run at the Olympic level and has continued for all subsequent Games. The Martin and ARRS marathon time lists differ primarily in several performances before WWIL, as there is some debate regarding the validity of early marathon times due to a lack of information regarding course-distance accuracy. Decision-making regarding the fastest marathon times on early marathon courses has always been arbitrary because the science of course measurement was only beginning to be defined. We therefore chose to analyze both FAP lists along with the WR list to see how the different data sets affected the prediction (see table 4 and figures 1, 2, and 3). The following equation was used:

Y= (Y, — Plateau)*exp(-K*X) + Plateau

Where Y, = the initial condition, Plateau is the theoretical fastest possible time, K is the rate constant, and X is performance time.

Using the WR data provides the best fit (r?=0.95, where the closer r? is to 1 the better the model is at predicting a trend), but the FAP lists fit nearly as well (r?=0.93 for ARRS and 0.92 for Martin). This was not surprising, as world-record progression data yields smaller error since record times continually decrease, making them more amenable to decay-function fitting. However, since these are best-of-the-best samples and discontinuous, the validity of the model may be affected when used for predictions. For example, it took 10 years for the 1988 record to be broken, but the 1998 record was broken three times in the next six years. Theoretically, the WR could be broken multiple times in a year or once in 20 years without the trend line itself being incorrect. However, for a sub-2:00 prediction, we are not interested so much in the trend line as a single point. With year-over-year data, the only error is on the y-axis as one sample must occur each

Table 4. Prediction models for marathon performance

FAP FAP WR Data (Martin data) (ARRS data)

Year of first sub-2 hour performance:

Fastest theoretical time possible:

2029 2032 2035

1:50:54 1:52:56 1:54:15

Best Fit Values

Yo 2.123E+16 7.338E+17 4.316e+017 K 0.01542 0.01726 0.01708 Plateau (sec) 6654 6776 6855

Pr 0.92 0.93 0.95

year. So in accordance with Liu and Schutz (1998), we conclude that the FAP model predictions are slightly superior to those using WR data.

Our models forecast the first sub-2:00 marathon performance occurring as early as 2029 and as late as 2035 with ultimate performances of 1:50:54 (Martin data), 1:52:56 (ARRS data), and 1:54:15 (WR data). Our results are similar to those performed by Peronnet and Thibault (1989), where a sub-2:00 marathon by the year 2028 and an ultimate performance of 1:48:25 were predicted using physiologically based models. Morton (1983) used time-versus-year data with exponential-decay models to predict running performances up to 5,000 meters that were later extrapolated to the marathon distance by Peronnet and Thibault (1989) to yield a projected ultimate performance time of 1:52:14. By comparison, Liu and Schutz’s (1998) exponential models predicted an ultimate performance of 2:00:43, never breaking through the two-hour barrier. Liu and Schutz also predicted a 2010 best time of 2:06:07 (actual was 2:04:48) and a 2050 best time of 2:02:39 (predicted to occur in 2012 to 2014 in our models). The underestimations of Liu’s models are likely due to the fact that more data is currently available (their predictions were made in 1998); thus, our models indicate that recent results have expanded the window of possible marathon times downward by about eight to 10 minutes. A summary of our results compared with past predictions is shown in table 5.

All three models predict modest yearly improvements in times, ranging from about seven seconds per year (WR data) to about 10 seconds per year (FAP—Martin data). This is what we would expect to see as the limits of human performance approach. So although we are merely 3 1/2 minutes away from the two-hour mark, our models predict it will take at least 17 years to get there. It should be noted that many new variables, plus the changing influence of existing ones, can potentially affect the rate of change, and as one can imagine, the number of contributing variables are extremely wide ranging—changing shoe technology,

Figure 1. Fastest Annual Performances – ARRS Data (No WWII)

Time (Seconds)

1900 1920 1940 1960 1980 2000 2020 Year

Figure 2. Fastest Annual Performances – AIMS Data (No WWII)

Time (Seconds) oc a So Oo

7000 1900 1920 1940 1960 1980 2000 2020

Year

population growth, performance-enhancing therapies, decimation of population due to war and/or disease, and more.

Table 6 on page 25 shows the actual versus predicted difference in our models for the last five world marathon records. Each of the data sets produced predictions that were within | percent of the actual performances.

With an idea of when the first marathon run under two hours may occur, it is also interesting to try to forecast who is likely to do it. Looking at the strongest marathon nations as assessed by the top 100 athletes and performances (by

Figure 3.

Time (Seconds)

1900

World Record Progression

1940 1960 1980 2000 2020 Year

Table 5. Performance predictions, past and present

Author (publication year)

Performance Projections

Lloyd (1966)

Ryder et al. (1976)

Rumball (1970)

Morton (1983)

Peronnet & Thibault (1989) Joyner (1991)

Liu & Schutz (1998)

Nevill & Whyte (2005) Present models

Martin FAP

ARRS FAP WR

assigning a point value of 100 to 1 and summing the totals for each nation) over the last five years can provide hints as to which country is most likely to lay claim to the first runner able to topple the barrier. The results, shown in table 7 on page 25, should not be a surprise to anyone following the sport—A frican nations, most notably Kenya, currently possess world dominance in the marathon, with one of the most impressive statistics being that 96 percent of the top 20 fastest yearly performances from the last five years have come from runners of African nationality. Using this data, it seems fairly safe to predict that the first sub-two-hour

2:02:00 (2000)

1:53:13 (2028)

1:37:30 (ultimate)

1:52:14 (ultimate)

1:59:36 (2028), 1:48:25 (ultimate)

1:57:58 (ultimate)

2:06:07 (2010), 2:02:39 (2050), 2:00:33 (ultimate) 2:03:38 (ultimate)

2:00:06 (2028), 1:57:27 (2050), 1:50:54 (ultimate) 2:00:33 (2028), 1:57:31 (2050), 1:52:56 (ultimate) 2:00:50 (2028), 1:58:11 (2050), 1:54:15 (ultimate)

Table 6. Actual WR versus predicted WR 2002-2011

Year Actual Athlete FAP – Martin Data FAP -ARRS Data WR Data Time (Nation) Predicted %Diff Predicted %Diff Predicted % Diff

2002.4 2:05:38 KhalidKhannouchi 2:04:40 0.8% 2:04:53 0.6% 2:04:20 1.0% (USA)

2003.7 2:04:55 Paul Tergat (KEN) 2:04:24 0.4% 2:04:37 0.2% 2:04:07 0.6%

2007.7 2:04:26 Haile Gebrselassie 2:03:36 0.7% 2:03:51 0.5% 2:03:28 0.8% (ETH)

2008.7 2:03:59 Haile Gebrselassie 2:03:24 0.5% 2:03:39 0.3% 2:03:18 0.6% (ETH)

2011.7 2:03:38 Patrick Makau (KEN) 2:02:50 0.7% 2:03:07 0.4% 2:02:51 0.6%

Table 7. Strongest marathon nations

2007 2008 2009 2010 2011

Nation Ath Pts _ Nation Ath Pts Nation Ath Pts Nation Ath Pts Nation Ath Pts KEN 65 3156 | KEN 68 3504 KEN 62 3333 KEN 58 2982 | KEN 65 3597

ETH 8 431 ETH 13 646 ETH 26 1113 ETH 29 1587) ETH 18 760 MAR 3 229 JPN 7 273 MAR 3 270 MAR 5 234 MAR 6 170 JPN 5 196 MAR 4 229 RSA 1 72 ERI} 1 58 ER} 2 121 CHN 2 = 129 ERI 2 107 UKR 1 57 UGA 1 52 USA 2 96

marathon performance will be run by a distance athlete from Africa, and if the trend toward the incredible Kenyan dominance continues, by one of its running stars. Can we go one step further? For those who really like to dream of knowing the future, we could go so far as to ask where the first sub-two-hour performance might occur. Table 8 shows the top 10 fastest marathon courses in the world, based on the average of the 10 fastest finish times. Based on the current data, Berlin is the most likely venue to produce the historic sub-2:00 feat, but over the next 20 years, new faster courses may arise or existing ones may be revised. As the athlete who achieves such a feat will be immortalized, so too will the race venue; therefore, it is likely that event organizers will continue to fine-tune the race environment as much as possible (such as large prize purses to attract the best runners and coordinated aids such as rabbits and fluid/carbohydrate delivery “on demand” from course vehicles) to help produce fast finish times and recordbreaking performances. It’s too much of a brazen stretch to try to predict the location of the first sub-two-hour accomplishment, but over the next few years, keep your calendars marked for the venues shown in table 8, as they are the most likely locations to produce world-record and/or annual-best marathon performances. Where do we go from here? What predictions can we make about the female marathon population? Is it possible for a woman to break two hours? Because the

women’s event is so much younger than the men’s (by 60 years in terms of Olympic participation), we unfortunately don’t have enough data on the women’s side to draw trend lines and make accurate predictions on future marathon performances. We do know, however, that a difference of about 10 percent separates the men’s times from the women’s across all standard running events from the 200 meters to the 42K distance, where an 8.7 percent time difference currently exists (Peronnet and Thibault 1989; Cheuvront et al. 2005). Given this, it is possible to estimate a theoretical fastest possible female marathon time using our current models (table 9).

In summary, our models predict that the first sub-two-hour male marathon performance will occur between 2029 and 2032, with a theoretical fastest possible time in the range of 1:50:54 to 1:52:56. Taking an audacious leap, we predict that this feat will likely be achieved by a runner from an African nation. We acknowledge that numerous mechanistic (flat course, cool temperature, cloudy skies, no wind, and so forth) as well as humanistic (excellent competition, ideal physiological attributes, optimally executed race tactics) details need to line up in an optimal manner for this milestone to be achieved. It’s anyone’s guess as to exactly when and where it will happen, so we’ll leave it up to the athletes to give us the answer.

Table 8. Fastest marathon courses*

Marathon Top 10 Avg Finish Times Event Month

1. Berlin 2:04:55 September

2. Rotterdam 2:05:07 April

3. London 2:05:21 April

4. Boston 2:05:39 April

5. Frankfurt 2:05:45 October

6. Chicago 2:05:58 October

7. Amsterdam 2:06:12 October

8. Paris 2:06:28 April

9. Fukuoka 2:06:47 December

10. Dubai 2:06:52 January *Through December 2011

Table 9. Ultimate marathon performance predictions—females

Martin AP Data —_ARRS AP Data WR Data

Ultimate performance 2:00:33 2:02:45 2:04:11 (@ 8.7% sex difference) Ultimate performance 2:01:59 2:04:14 2:05:40

(@ 10% sex difference)

Table 1. Annual fastest marathon performances by men (Martin data)

Year Athlete (Nation) Time Venue

1924 Shizo Kanaguri (JPN) 2:36:10 Tokyo, JPN

1925 Albert Michelsen (USA) 2:29:02 Port Chester, USA 1926 livari ROtk6 (FIN) 2:34:26 Helsinki, FIN

1927 Albert Michelsen (USA) 2:31:11 Port Chester, USA 1928 Boughera El Ouafi (MAR) 2:32:57 Amsterdam, NED 1929 Harry Payne (GBR) 2:30:58 Stamford Bridge, ENG 1930 Clarence DeMar (USA) 2:34:45 Los Angeles, USA 1931 Albert Michelsen (USA) 2:32:35 West New York, USA 1932 Tanji Yahagi (JPN) 2:31:31 Tokyo, JPN

1933 Kozo Kusunoki (JPN) 2:31:10 Tokyo, JPN

1934 Patrick Dengis (USA) 2:31:30 Port Chester, USA 1935 Kitei Son (JPN) 2:26:42 Tokyo, JPN

1936 Kitei Son (JPN) 2:28:32 Tokyo, JPN

1937 Manuel Dias (POR) 2:30:38 Lisboa, POR

1938 Patrick Dengis (USA) 2:30:28 Salisbury, USA 1939 Go Toyu (KOR) 2:31:26 Tokyo, JPN

1940 Ellison Brown (USA) 2:27:30 Salisbury, USA 1941 Salvatore Costantino (ITA) 2:32:33 Padua, ITA

1942 Zaiten Kimoto (JPN) 2:31:38 Tokyo, JPN

1943 Clayton Farrar (USA) 2:39:30 Yonkers, USA 1944 Donald Heinicke (USA) 2:45:05 Yonkers, USA 1945 Sven Hakansson (SWE) 2:36:37 G6teborg, SWE 1946 Mikko Hietanen (FIN) 2:31:37 Vuoksenniska, FIN 1947 Mikko Hietanen (FIN) 2:30:58 Loughborough ENG 1948 Mikko Hietanen (FIN) 2:31:02 Vuoksenniska, FIN 1949 Salomon Komonen (FIN) 2:28:40 Turku, FIN

1950 Feodosiy Vanin (URS) 2:29:10 Moskva, URS

1951 Veikko Karvonen FIN) 2:28:08 Tampere, FIN 1952 James Peters (GBR) 2:20:43 Chiswick, ENG 1953 James Peters (GBR) 2:18:35 Turku, FIN

1954 James Peters (GBR) 2:17:40 Chiswick, ENG 1955 Veikko Karvonen (FIN) 2:21:22 Kobenhavn, DEN 1956 Paavo Kotila (FIN) 2:18:05 Pieksamaki, FIN 1957 Sergey Popov (URS) 2:19:50 Moskva, URS 1958 Sergey Popov (URS) 2:15:17 Stockholm, SWE 1959 Sergey Popov (URS) 2:17:46 Kosice, SVK

Table 1. Continued

Year Athlete (Nation) Time Venue

1960 Abebe Bikila (ETH) 2:15:17 Roma, ITA

1961 Takayuki Nakao (JPN) 2:18:54 Nagoya, JPN 1962 Yu-Mang Hyang (PRK) 2:16:10 Pyongyang, PRK 1963 Leonard Edelen (USA) 2:14:28 Chiswick, ENG 1964 Abebe Bikila (ETH) 2:12:12 okyo, JPN

1965 Morio Shigematsu (JPN) 2:12:00 Chiswick, ENG 1966 Alastair Wood (GBR) 2:13:45 Forres, ENG 1967 Derek Clayton (AUS) 2:09:37 Fukuoka, JPN 1968 William Adcocks (GBR) 2:10:48 Fukuoka, JPN 1969 Derek Clayton (AUS) 2:08:34 Antwerpen, BEL 1970 Ronald Hill (GBR) 2:09:28 Edinburgh, ENG 1971 Derek Clayton (AUS) 2:11:09 Hobart, AUS 1972 Frank Shorter (USA) 2:10:30 Fukuoka, JPN 1973 John Farrington (GBR) 2:11:13 Richmond, ENG 1974 lan Thompson (GBR) 2:09:12 Christchurch, NZL 1975 Bill Rodgers (USA) 2:09:56 Boston, USA 1976 Waldemar Cierpinski (GER) 2:09:55 Montreal, CAN 1977 Bill Rodgers (USA) 2:10:56 Fukuoka, JPN 1978 Shigeru Soh (JPN) 2:09:06 Oita, JPN

1979 Bill Rodgers (USA) 2:09:28 Boston, USA 1980 Gerard Nijboer (NED) 2:09:01 Amsterdam, NED 1981 Rob de Castella (AUS) 2:08:18 Fukuoka, JPN 1982 Alberto Salazar (USA) 2:08:52 Boston, USA 1983 Rob de Castella (AUS) 2:07:51 Boston, USA 1984 Steve Jones (GBR) 2:08:05 Chicago, USA 1985 Carlos Lopes (POR) 2:07:12 Rotterdam, NED 1986 Taisuke Kodama (JPN) 2:07:35 Beijing, CHN 1987 Takeyuki Nakayama (JPN) 2:08:18 Fukuoka, JPN 1988 Belayneh Dinsamo (ETH) 2:06:50 Rotterdam, NED 1989 Juma Ikangaa (TAN) 2:08:01 New York, USA 1990 Stephen Moneghetti (AUS) 2:08:16 Berlin, FRG

1991 Koichi Morishita (JPN) 2:08:53 Oita, JPN

1992 David Tsebe (RSA) 2:08:07 Berlin, GER 1993 Dionicio Ceron (MEX) 2:08:51 Fukuoka, JPN 1994 Cosmas Ndeti (KEN) 2:07:15 Boston, USA 1995 Samuel Lelei (KEN) 2:07:02 Berlin, GER 1996 Martin Fiz (ESP) 2:08:25 Kyongju, KOR

Table 1. Continued

Year Athlete (Nation) Time Venue

1997 Khalid Knannouchi (MAR) 2:07:10 Chicago, USA

1998 Ronaldo da Costa (BRA) 2:06:05 Berlin, GER 1999 Khalid Knannouchi (MAR) 2:05:42 Chicago, USA 2000 Antonio Pinto (POR) 2:06:36 London, ENG 2001 Josephat Kiprono (KEN) 2:06:50 Rotterdam, NED 2002 Khalid Khannouchi (USA) 2:05:38 London, ENG 2003 Paul Tergat (KEN) 2:04:55 Berlin, GER 2004 Felix Limo (KEN) 2:06:14 Rotterdam, NED

2005 Haile Gebrselassie (ETH 2006 Haile Gebrselassie (ETH 2007 Haile Gebrselassie (ETH 2008 Haile Gebrselassie (ETH

2:06:20 Amsterdam, NED 2:05:56 Berlin, GER 2:04:26 Berlin, GER 2:03:59 Berlin, GER

2009 Duncan Kibet (KEN) 2:04:27 Rotterdam, NED 2010 Patrick Makau (KEN) 2:04:48 Rotterdam, NED 2011 Geoffrey Mutai (KEN) 2:03:02 Boston, USA

—

fee IS.neRy, f

MARAT realr | iw Sern = FCAl- Ee real oe

o e ¢

z gS

z & S = e

A Haile Gebrselassie breaks the tape at the 2008 Berlin Marathon in a then world-best of 2:03:59.

Table 2. Annual fastest marathon performances by men (ARRS data)

Year Athlete (Nation) Time Venue

1924 Shizo Kanaguri (JPN) 2:36:10 Tokyo, JPN

1925 Samuel Ferris (GBR) 2:35:58 Windsor, ENG 1926 livari ROtk6 (FIN) 2:34:25 Helsinki, FIN

1927 Verner Laaksonen (FIN) 2:35:21 Helsinki, FIN

1928 Boughera El Ouafi (FRA) 2:32:57 Amsterdam, NED 1929 Harry Payne (GBR) 2:30:58 London, ENG 1930 Fukutaro Shibui (JPN) 2:36:33 Tokyo, JPN

1931 Juan Zabala (ARG) 2:33:19 Kosice, SVK

1932 Tanji Yahagi (JPN) 2:31:31 Tokyo, JPN

1933 Kozo Kusunoki (JPN) 2:31:10 Tokyo, JPN

1934 Tamao Shiaku (JPN) 2:32:56 Osaka, JPN

1935 Kee-Jung Sohn (KOR) 2:26:14 Tokyo, JPN

1936 Kee-Jung Sohn (KOR) 2:28:32 Tokyo, JPN

1937 Manuel Dias (POR) 2:30:38 Lisbon, POR

1938 Pat Dengis (USA) 2:30:28 Salisbury Beach, USA 1939 Go Toyu (KOR) 2:31:26 Tokyo, JPN

1940 Shoichiro Takenaka (JPN) 2:33:42 Tokyo, JPN

1941 Leslie Pawson (USA) 2:31:27 Salisbury Beach, USA 1942 Zaiten Kimoto (JPN) 2:31:38 Tokyo, JPN

1943 Gérard Coté (CAN) 2:38:35 Yonkers, USA 1944 Charles Robbins (USA) 2:40:49 Yonkers, USA 1945 Sven Hakansson (SWE) 2:36:38 G6teborg, SWE 1946 Mikko Hietanen (FIN) 2:31:37 Vuoksenniska, FIN 1947 Mikko Hietanen (FIN) 2:30:58 Loughborough, ENG 1948 Mikko Hietanen (FIN) 2:31:02 Stockholm, SWE 1949 Salomon K6nénen (FIN) 2:28:39 Turku, FIN

1950 Fyeodosiy Vanin (SOV) 2:29:09 Moskva, URS

1951 Veikko Karvonen (FIN) 2:28:07 Tampere, FIN 1952 James Peters (GBR) 2:20:42 Chiswick, ENG 1953 James Peters (GBR) 2:18:35 Turku, FIN

1954 James Peters (GBR) 2:17:39 Chiswick, ENG 1955 Veikko Karvonen (FIN) 2:21:22 Kobenhavn, DEN 1956 Paavo Kotila (FIN) 2:18:05 Pieksamaki, FIN 1957 Sergey Popov (RUS) 2:19:50 Moskva, URS 1958 Sergey Popov (RUS) 2:15:18 Stockholm, SWE 1959 Sergey Popov (RUS) 2:17:45 Kosice, SVK

Table 2. Continued

Year Athlete (Nation) Time Venue

1960 Abebe Bikila (ETH) 2:15:16 Roma, ITA

1961 Takayuki Nakao (JPN) 2:18:54 Nagoya, JPN 1962 Mang-Hyang Yu (PRK) 2:16:10 Pyongyang, PRK 1963 Leonard Edelen (USA) 2:14:28 Chiswick, ENG 1964 Abebe Bikila (ETH) 2:12:11 Tokyo, JPN

1965 Morio Shigematsu (JPN) 2:12:00 Chiswick, ENG 1966 Alastair Wood (GBR) 2:13:45 Forres, ENG 1967 Derek Clayton (AUS) 2:09:36 Fukuoka, JPN 1968 William Adcocks (GBR) 2:10:48 Fukuoka, JPN 1969 William Adcocks (GBR) 2:11:07 Athens, GRE 1970 Ron Hill (GBR) 2:09:29 Edinburgh, SCO 1971 Derek Clayton (AUS) 2:11:09 Hobart, AUS 1972 Frank Shorter (USA) 2:10:30 Fukuoka, JPN 1973 John Farrington (AUS) 2:11:13 Sydney, AUS 1974 lan Thompson (GBR) 2:09:12 Christchurch, NZL 1975 Bill Rodgers (USA) 2:09:56 Boston, USA 1976 Waldemar Cierpinski (GER) 2:09:55 Montreal, CAN 1977 Bill Rodgers (USA) 2:10:55 Fukuoka, JPN 1978 Shigeru So (JPN) 2:09:06 Beppu, JPN 1979 Bill Rodgers (USA) 2:09:28 Boston, USA 1980 Gerard Nijboer (NED) 2:09:01 Amsterdam, NED 1981 Rob de Castella (AUS) 2:08:18 Fukuoka, JPN 1982 Alberto Salazar (USA) 2:08:52 Boston, USA 1983 Rob de Castella (AUS) 2:08:37 Rotterdam, NED 1984 Steve Jones (GBR) 2:08:05 Chicago, USA 1985 Carlos Lopes (POR) 2:07:12 Rotterdam, NED 1986 Rob de Castella (AUS) 2:07:51 Boston, USA 1987 Takeyuki Nakayama (JPN) 2:08:18 Fukuoka, JPN 1988 Belayneh Dinsamo (ETH) 2:06:50 Rotterdam, NED 1989 Juma Ikangaa (TAN) 2:08:01 New York, USA 1990 Steve Moneghetti (AUS) 2:08:16 Berlin, GER

1991 Koichi Morishita (JPN) 2:08:53 Beppu, JPN 1992 David Tsebe (RSA) 2:08:07 Berlin, GER

1993 Dionicio Ceron (MEX) 2:08:51 Fukuoka, JPN 1994 Cosmas Ndeti (KEN) 2:07:15 Boston, USA 1995 Sammy Lelei (KEN) 2:07:02 Berlin, GER

Table 2. Continued

Year Athlete (Nation) Time Venue

1996 Martin Fiz (ESP) 2:08:25 Kyong-Ju, KOR 1997 Khalid Khannouchi (MAR) 2:07:10 Chicago, USA 1998 Ronaldo da Costa (BRA) 2:06:05 Berlin GER

1999 Khalid Khannouchi (MAR) 2:05:42 Chicago, USA 2000 Antonio Pinto (POR) 2:06:36 London, ENG 2001 Josephat Kiprono (KEN) 2:06:50 Rotterdam, NED 2002 Khalid Khannouchi (USA) 2:05:38 London, ENG 2003 Paul Tergat (KEN) 2:04:55 Berlin, GER 2004 Evans Rutto (KEN) 2:06:16 Chicago, USA 2005 Haile Gebrselassie (ETH) 2:06:20 Amsterdam, NED 2006 Haile Gebrselassie (ETH) 2:05:56 Berlin, GER 2007 Haile Gebrselassie (ETH) 2:04:26 Berlin, GER 2008 Haile Gebrselassie (ETH) 2:03:58 Berlin, GER 2009 Duncan Kibet (KEN) 2:04:26 Rotterdam, NED 2010 Patrick Makau (KEN) 2:04:48 Rotterdam, NED 2011 Geoffrey Mutai (KEN) 2:03:02 Boston, USA

g s Qo

A Geoffrey Mutai runs a 2:03:02 at the 2011 Boston Marathon.

Table 3. Marathon world record progression (IAAF data)

Year Time Athlete Nation Venue 1925 2:29:01 Albert Michelson USA Port Chester, USA 1935 2:27:49 Fushashige Suzuki JPN okyo, JPN 1935 2:26:44 Yashuo Ikenaka JPN okyo, JPN 1935 2:26:42 Kitei Son JPN okyo, JPN 1947 2:25:39 Yun Bok Suh KOR Boston, USA 1952 2:20:42 Jim Peters GBR Chiswick, ENG 1953 2:18:40 Jim Peters GBR Chiswick, ENG 1953 2:18:34 Jim Peters GBR urku, FIN 1954 2:17:39 Jim Peters GBR Chiswick, ENG 1958 2:15:17 Sergey Popov URS Stockholm, SWE 1960 2:15:16 Abebe Bikila ETH Rome, ITA 1963 2:15:15 Toru Terasawa JPN Beppu, JPN 1963 2:14:28 Leonard Edelen USA Chiswick, ENG 1964 2:13:55 Basil Heatley GBR Chiswick, ENG 1964 2:12:11 Abebe Bikila ETH okyo, JPN 1965 2:12:00 Morio Shigematsu JPN Chiswick, ENG 1967 2:09:36 Derek Clayton AUS Fukuoka, JPN 1969 2:08:34 Derek Clayton AUS Antwerp, BEL 1981 2:08:18 Rob de Castella AUS Fukuoka, JPN 1984 2:08:05 Steve Jones GBR Chicago, USA 1985 2:07:12 Carlos Lopes POR Rotterdam, NED 1988 2:06:50 Belayneh Dinsamo ETH Rotterdam, NED 1988 2:06:05 Ronaldo da Costa BRA Berlin, GER 1999 2:05:42 Khalid Khannouchi MAR Chicago, USA 2002 2:05:38 Khalid Khannouchi USA London, ENG 2003 2:04:55 Paul Tergat KEN Berlin, GER 2007 2:04:26 Haile Gebrselassie ETH Berlin, GER 2008 2:03:59 Haile Gebrselassie ETH Berlin, GER 2011 2:03:38 Patrick Makau KEN Berlin, GER References

Cheuvront, S. N., R. Carter, K.C. DeRuisseau, and R. J. Moffatt. 2005. Running performance differences between men and women: an update. Sports Medicine 35(12):1017-1024.

Henry, F. M. 1954. Prediction of world records in running sixty yards to twenty-six miles. Research Quarterly 26:147-158.

Joyner, M. J. 1991. Modeling: optimal marathon performance on the basis of physiological factors. Journal of Applied Physiology 70(2):683-687.