Saturday, 24 August 2013

Ultramarathon Running Research Project - Thames Path 100 2013

This year, I have continued the project that I started last year with James Elson from Centurion Running, looking at various aspects of ultra running and how these relate to a runner's ability to complete the race. I still think that there is lots of interesting information to be mined from these data, and hopefully more people will take part in the surveys for the next few races. Anyway, below is the report that I produced for the 2013 Thames Path 100 mile race in its entirety. As ever, these are merely my own interpretations of the data, but I would love to hear from anyone that might have any alternative ideas. I hope that you find it interesting!


Following on from last year's pilot project at the South Downs Way 100 mile race (SDW100) last year, we have now upgraded the analysis to look at all four of the Centurion Running 100 mile events in 2013. The current state of research into the factors that may affect a runner's ability to complete a 100 mile event is still very open, with several key studies beginning to delve into the key factors essential for all runners to consider.

The research of Martin Hoffman and the rest of the Western States Endurance Run Research Committee has produced several key papers, analysing both the effects of running 100 miles on the body, as well as the demographics of the runners who choose to run such events. Recently, the Ultrarunners Longitudinal TRAcking (ULTRA) study from Stanford University (which I recommend everybody takes part in if they haven't already by going to has begun to track a wide range of information for ultrarunners, which will then be tracked over the coming years with the hope of monitoring training styles and observing how these relate to injury rates.

Recently, more and more researchers have aligned themselves with race directors in order to obtain data on ultrarunners in the field. For many races, this may allow the study not only of runners' preferences and training styles, but also the study of the changes in physiology brought about as a result of running 100 miles. These studies will undoubtedly direct our current understanding of "optimum" training methods in order to ensure that running styles are tailored to specific physiological needs. Whilst I personally do not believe that there is such a thing as "the right way to train" (it is likely highly dependent on individuals), there are certainly universal truths that we can all benefit from fully understanding.

Our own study was a fairly simple yet powerful approach. We asked runners of the SDW100 to complete a pre-race survey (focussing on information on the runners themselves and normal training strategies) and a post-race survey (focussing on their approach to the race and how the race itself went for them), and combined these data with split times throughout the race. There were several goals with these data, but the main goals were to understand what sort of people typically take part in such events, what sort of training strategies are typically used, and how these relate to race-day performance.

In 2013, we will be performing these analyses on all four of the Centurion Running 100 mile races; the Thames Path 100, the South Downs Way 100, the North Downs Way 100, and the Winter 100. At the end of the year, we will combine these data into a single analysis to get a view of how runners have approached the different races through the year. These analyses are only possible because of volunteers choosing to take part in the surveys. But the more people who take part, the more interesting the findings will be. I hope that these results will encourage people to take part in the future studies and that we will soon have a deep pool of data to mine for interesting results.

On a personal note, my time for performing these analyses has been drastically reduced due to the birth of my gorgeous little girl. She's strong on her feet already so I'm sure she'll be a runner like Daddy! But unfortunately she also seems to have my lung capacity and ability to cope with no sleep... But I hope that you find what I have managed to cobble together between feeds interesting! These are entire my own opinions, but if you have any thoughts or comments feel free to contact me through my blog at As ever, all the best to everybody with your running, and I'll see you out on the trails!

The Race

The Thames Path 100 (TP100) is a 100 mile non-stop race on foot along the River Thames from Richmond, London to Oxford with a 30 hour time limit. This race is perhaps one of the flattest ultramarathon routes in the world, with a total climb of only 2,500 feet. The current course record stands at 15:11:15 from Craig Holgate in 2012.

Due to very heavy rain surrounding the event, the course was considerably adjusted to avoid flooded areas around the river. The race ran from Richmond to Windsor, before heading out on several out and back sections; between Windsor and Cookham, between Windsor and Walton, then between Windsor and Cookham, finishing at the main Centurion Running base in Windsor. This course was in fact slightly longer than 100 miles, and for this reason, it is not entirely comparable to the inaugural 2012 event.

The race was also hit by heavy rain and snow making conditions incredibly tough for all runners. This is highlighted in the finishing rate of what is, ostensibly at least, one of the "easier" 100 mile courses. Of the 165 runners (20 female and 145 male) who started the 2013 event, only 90 (54.55 %) completed the entire course. This clearly shows that there is no way to predict the British weather, and also that there is no such thing as an "easy" 100 mile race!

Of these runners, 41 (24.85 % of the starters) opted to take part in our survey (8 female and 33 male). The finishing rate for those taking part in the survey was 30 finishers, giving a finishing rate of 73.17 % amongst those that took part. This suggests a slight bias for people who actually finished the race to take part in the survey.

There were 11 checkpoints along the way:

Checkpoint Name
Walton 1
Wraysbury 1
Windsor 1
Cookham 1
Windsor 2
Wraysbury 2
Walton 2
Wraysbury 3
Windsor 3
Cookham 2
Windsor 4



Pacing a 100 mile run remains a complex and controversial aspect of ultra running. Whilst most people would advocate attempting to run an even pace throughout the race, others would suggest that runners should try to maintain a constant feeling of effort throughout the race. Others still, such as the winner of the 2013 Lakeland 100 Stuart Mills, prefer to run using the motto "run as hard as you can for as long as you can", and then just hanging on until the finish. In all honesty, as with most aspects of ultra running, I would be surprised if there existed one single optimum pacing strategy. There are simply too many factors involved. And it is pretty clear from watching the top runners in many races that different strategies work for different people.

Regardless, it is incredibly useful to look at the overview of runners' paces throughout the race to understand how different runners approached pacing. Figure 1 shows the distribution of speeds (in miles per hour) between the 11 checkpoints for all runners who completed the race. The speed is calculated based on the distance between successive checkpoints and the time taken between them. Male (blue) and female (red) runners are shown separately. The white dot in the center of each "violin" represents the median value (the middle value if you rank them from lowest to highest), and the width of the violin represents the proportion of runners showing that speed. So essentially, most runners ran at the speed where the violin is fattest.

Figure 1: Distribution of speed (miles per hour) of all runners between checkpoints
In the previous analysis for the 2012 SDW100 race, it was clear which were the toughest sections of the race based on sudden drops in the speed of runners over certain sections. However, in this case there are no such sections. This is perhaps not surprising given the complete lack of hills along this route. However, what is clear is that runners gradually got much slower as the race progressed. Probably as a result of the flatter course, all runners started out faster than on the 2012 SDW100, but by the end runners had slowed down significantly (largely as a result of the terrible conditions on the day). Another interesting comparison with the SDW100 results is that we no longer see a significant difference between men and women. With the SDW100 analysis, women started at what seemed to be a slower, or perhaps more sensible pace than the men, and thus their drop in pace over the race was less severe. However, in the TP100 this is no longer the case. Whether this indicates a more competitive field, or whether all runners were lulled into a false sense of security by the flat course, I cannot say.

Figure 2 shows the same data, but this time shows the pacing profile for the individual runners by connecting the points between the aid stations with a straight line. The top 5 male finishers (blue) and the top 5 female finishers (red) are highlighted. The colour of the lines has been graded such that the top finishers profile has the darkest colour. Here we can see that most of the runners ran a pretty even race throughout the event (with no sudden drops in speed). There were two runners who took off at a much faster pace than the other runners (Martin Bacon who ultimately won the race, and David Ross who came 6th), but ultimately slowed to a more consistent pace.

It is interesting to compare the pacing strategies of the first and second place male runners. As mentioned above, Martin Bacon's approach was to start at a relatively fast pace of about 7:30 mins/mile, ultimately slowing significantly from the 28 mile Windsor checkpoint to almost 11:30 mins/mile, and gradually slowing to a final pace of about 13:30 mins/mile by the end of the race. In comparison, second place runner Luke Ashton ran a much more even-paced race, starting at a comparatively sedate pace of about 8:30 mins/mile, and slowing gradually throughout the race to a pace of about 10:00 mins/mile for the majority of the race, and finishing at a similar final pace to Martin of 13:30 mins/mile. So they both ran vastly different races, and yet the final finishing times were 18:10:53 for Martin, and 18:14:18 for Luke - only three and a half minutes in it! So whilst Martin's approach had him far out head in the first third of the race, the cost to him was to slow significantly allowing Luke's more consistent pace to catch up with him. It was a tortoise and hare situation, but the hare managed to hold on until the end.

The results of the women's race were more pronounced, with Debbie Martin-Consani running an amazing race to finish first lady (and fourth overall) in 19:19:20. Wendy Shaw came second female (and tenth overall) in a time of 20:58:15. So whilst Debbie was only an hour behind the overall winner, Wendy was over an hour and a half behind her. Both runners followed a similar approach; starting out at a relatively sustainable pace, and slowing gradually over the course.

I think that these results fit with the thesis that there is no one way to run an ultra. The choice of pacing strategy is largely going to come down to you as an individual. What works best for you? Chances are you will only find this out by trying different things in training. There are certainly objective pros and cons to each approach: running fast at the start allows you to break away from the main pack and allows you to make up ground while you are feeling fresh, but has the effect of thoroughly trashing you early in the race; running consistently allows you to hold onto energy reserves and protect your muscles, but you can get snarled up in groups of runners and held up at checkpoints. I believe that Martin's win goes to show that - perhaps despite conventional wisdom - for some runners the approach of going out strong can be a great strategy. However, Luke's almost equally amazing run goes to show that a completely different strategy can, for the right runner, be equally good.

So rather than show the optimum running strategy, this further adds to confusion showing that there probably is no best way to run an ultra.
Figure 2: Pacing profile of the top 5 male (blue) and female (red) runners


It is an unfortunate fact of ultra-marathons that not everybody will finish the race. A runner may be forced to pull out due to any number of reasons. These can relate to insufficient training, but sometimes things just don't go right on the day. A lot can go wrong over the time it takes to run 100 miles. In fact, it is almost inevitable that something will go wrong along the way. In many cases, mental fortitude is sufficient to drive through these times (how many times have you been flagging at 70 miles thinking you were going to die, only to be running like the wind 10 miles later?), but other times for whatever reason a runner may decide to pull out from the race. This may be because of an injury or some other problem that they do not want to make worse by pushing on, or may be due to feeling low with no sign of improvement and a long way still to go. Whilst 100 mile running is clearly a physical activity, requiring a lot of training to achieve, it is clear that there is also a large mental element to running. Different people have different ways of coping with the bad times in ultramarathons, but there is no doubt that a) 100 miles is a long way to run, and b) it will probably hurt!

Figure 3 shows the number of runners who dropped out at each of the 11 checkpoints along the course. The height of each bar indicates the number of runners who dropped out at that aid station for male (blue) and female (red) runners. The number at the top of each bar represents the cummulative number of DNFs up to that point.

Figure 3: Number of runners who dropped out at each aid station
The conditions at the TP100 were particularly bad, and so the DNF rate was particularly high, with 75 out of 165 runners (45.5 %) dropping before the finish line. The rerouted course was almost a cross-shape centered at Windsor, with runners returning to the main Windsor checkpoint 4 times throughout the race. This checkpoint was also where runners' drop bags were located, along with warm food and cover from the rain and snow. It therefore was clearly a tempting point at which to stop running, such that forcing yourself to head away from such (relative) comforts may be difficult. Of course the corollary to this point is that the design of the course meant that you were never too far from the safety of the Windsor checkpoint.

The first checkpoint where we see a significant increase in DNFs was the first time that runners reached the Windsor checkpoint. This was both the first time that runners reached the larger checkpoint where all of their equipment was stored, but also was the first checkpoint past the marathon distance of 26.2 miles. Many people likely ascribe significance to certain distances in these races, with the marathon distance representing the first. One could imagine that running a tough race such as this under such difficult conditions, somebody might think to themselves, "I'll at least run a marathon".

The number of DNFs is higher at the next aid station at Cookham (38 miles), possibly representing people running past the marathon distance for the first time, but unfortunately running into issues.

The checkpoint with the highest dropout rate was the return to Windsor at 48 miles. My interpretation is that this represents another milestone distance in the race (about halfway), where runners have to decide whether or not they can possibly run the same distance again in order to finish. This is an interesting mental challenge, and I personally ascribe to the idea of never thinking about the distance as a whole, and instead trying (it doesn't always work...) to only think from checkpoint to checkpoint. But if you arrive at the midway point feeling terrible, it is easy to see how the idea of running the same again would not be entirely appealing.

From this point, whilst the DNF rate remains quite high, it also declines over the next few checkpoints. However, the Wraysbury checkpoint at 76 miles shows a spike in DNFs, perhaps representing another milestone in the race (three quarters of the way). By this point, runners have made a huge dent in the race, but there is still a marathon remaining. [Edit: It has been pointed out to me that this aid station was the only one indoors, so represented a nice comfy place to stop!]

However, after this point, the DNF rate drops off significantly, with only a small number dropping out in the last 24 miles. This probably reflects runners refusing to drop out so close to the finish, instead choosing to push on through to the finish. I am particularly impressed by the low dropout rate at the 82 mile Windsor checkpoint. Forcing yourself out back into the wet and muddy trail for another 18 miles when you are already at the finish line must be incredibly tough! This is the inherent difficulty with a race format like this (similar to a looped course). It is all too easy to get comfortable at the home-base.

Of course, these are purely my own interpretations of these results, but the same sorts of milestones were identified in the four Centurion Running races last year as well. I would be interested to hear if anyone out there does this - ascribing particular significance to specific points in the race, and thinking in overall terms of the distances. Or do you prefer to focus on the immediate task at hand and break the race down into smaller more manageable chunks? I like to "run stupid" and try not to think too much about what's coming, but that's just me.


The way that runners prepare for a 100 mile race is another aspect of running open to debate. Should you taper before the race? If so, for how long? How far should your longest training run be? How far out from the race should you run your longest training run? How many miles should you run each week? As with many aspects, these factors are likely to be very much a personal preference. I, for instance, don't like to taper, and will usually continue my usual mileage pretty much up to race day. The figures below show various aspects of training leading up to race day, which may influence a runner's performance.

Figure 4 shows the perceived difficulty of the race, on a scale of 1 (easy) to 10 (difficult). The majority of runners found the race to be incredibly difficult, with an average score of 8/10. Whilst there was no difference in the overall perceived difficulty for finishers and non-finishers, almost all non-finishers rated the difficulty as 7 or higher.

Figure 4: Perceived difficulty of the race from runners who took part in our survey. Runners rated the difficulty of the race on a scale of 1 (easy) to 10 (difficult).
Figure 5 shows the distribution of the highest maximum mileage run by competitors in the 2 months prior to the race. Most runners tend to run around 50-70 miles per week, and there does not appear to be any difference between runners who finished the race vs. those that did not. It appears that the female competitors typically run more miles on a weekly basis than male competitors, although this is not a statistically significant increase, and the number of females who took part in the survey was quite low. However, this may suggest that women typically prefer to feel ready and well trained before entering a race such as this, which may also help to explain the much lower levels of women taking part in 100 mile races.

Figure 5: Distribution of the highest weekly mileage of runners in the 2 months leading up to the race for runners who took part in our survey
Figure 6 shows the distribution of the longest training run in the 2 months before the race for all runners who took part in our survey. There appears to be a pretty striking cutoff of 50 miles for most runners, with only a small number of runners having run further than this prior to the race. The average for men is around 35 miles, and is over 40 miles for women. Such distances make sense as they allow runners to significantly break the psychological barrier of the marathon without actually having to run the full distance before the day. This fits in with the idea that ultra marathons are "95% mental; the other 5% is all in the mind". It is interesting that the average longest run for women is higher than for men. This may again suggest that female runners only enter events such as these when they feel well trained for it, but the sample size is too low to draw any real conclusions. Interestingly, the runners who didn't finish the race generally ran longer training runs prior to the race, although the sample size is quite small.

Figure 6: Distribution of the longest training run of competitors in the 2 months prior to the race
Figure 7 shows the distribution of the number of years for which runners have been taking part in ultramarathons. To many people, ultrarunning is a relatively new sport, so it is really interesting to see runners who have been competing for 30 years or more in such events. Also interesting is that these runners, despite being amongst the older competitors, all finished the race. There is no substitute for experience! Also interesting is that female runners have typically been running for longer than male runners, again suggesting that perhaps women runners prefer not to take part in such events until they feel fully ready and have the experience to back up their attempts.

Figure 7: Distribution of the number of years for which participants in our survey have been running ultra marathons

Runner Characteristics

"What sort of an idiot would want to run 100 miles"?! This is a question that we have all likely faced when people find out what we enjoy doing in our spare time. But the answer to this simple question might just surprise people not already in the know. I recently had a conversation with my mother where she said that she had always assumed that these kinds of races were the domain only of the ├╝ber-fit, yet the reality is that trail and ultra running truly is for everybody. Whilst it takes a huge amount of commitment and training to complete a 100 mile race, ultimately I truly believe that anybody can do it. I also believe that many ultra runners are like me, in that one of the driving forces for their running is the ability to eat lots of cakes along the way, and drink beer at the end!

Figure 8 shows the distribution of males and females amongst the entire starting field, and how many of each group completed the race. Sadly it seems that there is a clear bias in the field for male runners, so there is plenty of room for more ladies in future races! The finishing rates were pretty similar for males and females, so women were just as likely to complete this gruelling course as the men - showing that there is no room for sexism in this sport!

Figure 8: Distribution of male and female competitors across the entire field
Figure 9 shows the distribution of the age of all competitors across the field for males and females. These data are represented as violin plots, with the width of each violin indicating the proportion of competitors with that age. The white dot in the middle of each violin represents the median value for that group. Any groups that are significantly different from one another are indicated with the corresponding p-value from a 2-sample Wilcoxon rank-sum test. One thing that is quite interesting here is that the women in the race are generally a little younger than the men, with the men being on average in their mid-40s, but with women being typically in their late 30s. This is even more striking when looking at the finishers, where this difference is even more pronounced. However, the sample size is really too small to make any firm assertions.

Figure 9: Distribution of the age of competitors across the entire field split by gender

The following figures (Figure 10 to Figure 18) are a selection of potentially interesting characteristics of the runners who opted to take part in our survey. These figures illustrate the wide variety of people who are drawn into the wonderful world of ultrarunning. Apparently having blonde hair and hazel eyes is a good way to ensure that you avoid a DNF! That, and avoiding meat in your diet. Oh, and apparently women are typically shorter than men. Who knew! Yay for science.

(Please note that, in case it's not obvious, I am being facetious - please don't go dyeing your hair and cutting meat out of your diet, then getting annoyed when it doesn't improve your running!)
Figure 10: Distribution of runners' weight class, based on their calculated BMI, for runners who took part in our survey (BMI < 18.5 - Underweight; 18.5 < BMI < 25 - Normal; 25 < BMI < 30 - Overweight; BMI > 30 - Obese)

Figure 11: Highest form of education for competitors who took part in our survey
Figure 12: Distribution of employment status for runners who took part in our survey
Figure 13: Eye colour for runners who took part in our survey

Figure 14: Relationship status for all runners who took part in our survey

Figure 15: Distribution of runner heights for all runners who took part in our survey

Figure 16: Religious views of runners who took part in our survey

Figure 17: Hair colour for runners who took part in our survey

Figure 18: Dietary preferences of runners who took part in our survey


As with the South Downs Way 100 analysis from last year, looking over these data has been quite interesting if not entirely clear-cut. The overall conclusion to draw from much of these data is that there really is no one way to run or prepare for a 100 mile ultramarathon. In many ways this was a slightly odd race to analyse, since I believe that the conditions on the day of the race (which were not just tough, but more importantly were much tougher than most runners had prepared for) likely had more of an effect on people's ability to complete the race than any aspect of their actual training.

There were, however, several interesting results which are summarised below:
  1. Anybody can run an ultramarathon. Runners of the TP100 came from all walks of life, with people of all ages, heights, weights, etc. present. Unfortunately there does seem to be a lack of women taking part in such races. But I truly believe that anybody can run a 100 mile race, providing they are willing to put in the effort to get there. Whether they would want to is a completely different question...
  2. Pacing of a 100 mile race is very much a personal choice, although all runners will slow significantly as the race progresses. Conventional wisdom suggests that heading out too fast is a sure-fire way to ensure a DNF, but it seems that this approach may work weel for some people. Whether you choose to run fast at the start and slow more as the race goes on, or aim for a more consistent pace throughout, may not necessarily indicate your overall performance. My advice to anybody running a 100 miler for the first time is to do it slowly with no time goals in mind, and just make sure that you make the finish line. But if you are racing, there is no reason to think that heading out fast is not a perfectly valid choice of strategies. My personal approach is to try to maintain a relatively constant effort level throughout the race, so that the start feels just as difficult as the end.
  3. I believe that it is easier to justify pulling out of the race at certain points, and our ability to push on can depend very much on how we weigh up the pain of continuing with the disappointment of pulling out. It is harder to consider pushing on when we still have 50 miles to go than it is when we are already 95 miles through the race. I personally have no problems with DNFing as long as I can justify to myself that my future-self won't be disappointed in me. For instance, I pulled out of the TP100 in 2012 at mile 78 as an injury that I had sustained the week before flared up. I was of course disappointed, but I wasn't willing to injure myself further for the sake of not DNFing. In comparison, I walked the last 10km of the 2012 UTMB like John Wayne due to some rather painful chaffage, but the pain of walking was outweighed by the chance to see the amazing sites along the way in a place I probably won't get the chance to visit again any time soon.
  4. 50 miles seems to be a good maximum training distance for runners considering taking part in a 100 mile race. If you can do 50 miles, I would say that you can do 100 miles.
  5. Most runners ran a maximum of 50-70 miles a week in training. This seems a pretty sensible and sustainable weekly mileage, and is in line to what I typically run.
  6. It appears that women who took part in this race were typically quite experience. The results of the SDW100 analysis suggested that women might be more sensible in taking part in 100 mile races, in as much as only those women who have been running for a while and regularly run long distances take part in. In comparison, men are more likely to jump straight in without building up to it. The results of the TP100 analysis seem to back this up, although the sample size for women runners is quite small. I will say that I count myself in this "silly men jumping straight into things without thinking them through" group, having run my first 100 mile run directly after my first marathon (I think I misunderstood the idea of a warm-down run...).
  7. This was a very, very tough race, despite being almost entirely flat.
There are ultimately many interesting factors in these data, and I hope that some of these results have been interesting. These are merely my own interpretations of things, and I would love to hear from you if you have any additional insights. If you are taking part in any of the other races in the series, I urge you to take part in the surveys to allow us to allow us to pull out even more interesting information. Next up is the South Downs Way 100 2013 survey, so watch this space!


  1. Gary Gellin, Menlo Oark, CA26 August 2013 16:13

    How do your tortoise and hare competitors who finished a close 1st and 2nd in the 100 compare at shorter distance races?

    1. Good question. Having a quick look around online (does that count as stalking?!), I see that Luke's best marathon time this year was 2:56:26 (Edinburgh Marathon), whilst Martin's was 3:01:41 at the Abingdon Marathon.