Race report: Running 100 kilometers in 6:54

So I ran 100 kilometers last weekend. Crazy idea?—I agree. Here, I want to describe how the challenge of running the distance in less than seven hours developed, became unrealistic during preparation and even more over the course of the race—and still ended up being achievable.

While I do not want to dive too much into the why at this point (I might write a separate post about it later), there are at least three different aspects to it: First of all, it was my first “real” foray into ultrarunning1, posing both an immense mental and physical personal challenge. Additionally, running 100 km on a flat course, as opposed to an ultratrail, is also an endeavor that yields objectively measurable and comparable finish times as a manifestation of individual limits. Well, and it’s 2020—without any other athletic events coming up, it seemed like a good time to move from an ambition in the back of the head towards an actual goal.

Is a sub-seven time possible?

While running 100 kilometers itself certainly would not be easy, I also wanted to run them fast. But what does “fast” on such a distance mean for me? After some brief online research to get ballpark estimates, it became clear that runners who had run faster than 6:50 usually had marathon PRs in the sub-2:20 range, while some of those with times around 7:00 had marathon PRs comparable to mine (2:26).

Given that seven hours form a (psychological) barrier, it quickly became an abstract goal—but not necessarily a realistic target: Looking at marathon PRs of seven-hour runners (rather than the other way around) is obviously biased.2 Instead, I focused on pacing data (10km splits from the flat courses of Winschoten and Los Alcazares) from a practical point of view: How likely is a sub-seven finish given the the starting pace (as a proxy for each runner’s goal)?

Finish times in relation to starting pace on the first 10km: The left plot shows that runners start out substantially faster than the required mean pace (dotted line). The right shows that the fraction of runners achieving sub-seven finishes is already less than 50% for runners starting out with a 4:00 min/km pace. I created the plots based on data of 113 finishers from Winschoten in 2011 and 2015 as well as Los Alcazares in 2016

Clearly, a sub-seven finish is not very likely, even for runners who start out substantially below 4 min/km. As a perspective, crossing the finish line in seven hours requires an average of 4:12 minutes per kilometers, while 4:00 min/km would yield 6:40. Note that even this is probably optimistic since it contains participants of the world championships which likely demonstrated strong performances before.

Overall, while a sub-seven finish seemed possible, it was far from expected.3 Nevertheless, the goal was attractive enough to start out at a seven-hour pace and accept the higher risk of a blow up in the later stages.

DIY-Racing: Benefits of control or paradox of choice?

Having committed myself to the project around early August, a race date in mid-October seemed reasonable.4 This left time for eight weeks of specific training plus a two week taper. I stayed true to my general approach of prioritizing quality over quantity in training, and adapting the plan according to perceived exertion. Over the training block, I averaged 117km per week (about twenty more than during a typical marathon training), accompanied by some easy cycling and swimming. Training went reasonably well, with three long runs between 50km and 55km.5

Assembling a crew, I was lucky that Janosch , with his experience and successes as a professional trailrunner, was immediately on-board. After his triumph in Innsbruck over the K110 in September, he understandably opted for keeping me company on the bike rather than running along. Yet this actually allowed me to profit from on-demand catering (of water and gels) and an appropriate race-state-dependent balance of encouragement and shouting. As additional crew, I was happy to motivate Christian, Felix, Karsten, Oliver, Patsy and Theo to run along, with individual legs between 5 and 50 km.

Choosing the course turned out to be more difficult. My search around Freiburg for a paved flat course with minimal intersections did not bear any fruits. Instead, the Alsace offered better prospects, with a 50km out-and-back course along the scenic Canal du Rhône au Rhin becoming the favorite. Yet given the evolving COVID situation I opted for a 25km loop in Karlsruhe’s Hardtwald, my reliable training environment during university.

Starting mindset: First go for a few training runs, the race will start early enough

There were two certain things before the start: Pacing is essential. And there will be suffering, no matter the pacing. While I would find out whether my persistence would be enough for the latter, pacing could actually be somewhat controlled and therefore planned—both quantitatively and mentally.

From a quantitative point of view, analyzing the median evolution of 10km-splits from three world championships (which I will write about separately) gave some objective guidance: Successful runners, managing to run sub-seven times, started out about 8 seconds faster per kilometer than overall mean pace, slowed down after 50 or 60 kilometers and ended up taking about 16 seconds/kilometer more than the mean on their final 10km. Given a target of 6:58 to account for short breaks, this meant a starting pace of about 4:03, slowing down to 4:27 min/km at the end.

More information on how I came up with this strategy as well as plots addressing common split times can be found in my post on pacing.

Mentally, I tried to copy my Ironman race strategies from Nice and Kona: Everything needs to feel easy until the final half-marathon. While the comparison is not really appropriate—no low heart rate and easy nutrition on the bike leg, no variation in muscular demands, potentially less feeling of progress—this led to the goal of a 50km prologue, followed by starting a long run on heavy legs, topped off by a half-marathon of hard racing.

The first 45 kilometers: Starting on cruise control

No starting corral, no gun shot, but still in a group of five runners: The start certainly felt different from any of my last major races, but not worse. Given the distance to go, the lack of an adrenaline rush was probably a good thing, and we still started out about 20 seconds too fast on the first kilometer, but settled quickly into a steady 4:00-pace, finishing the first 10km in slightly less than 40 minutes.

With Theo running along for the first 25km and Olli staying around for 12 more, it was easy to focus on maintaining constant intensity and avoiding any potential muscular risk such as uneven parts of the road. While running alone after the first two-and-a-half laps, the legs were still running smoothly enough that I was trying to extrapolate how much time might be lost due to the headwind on the straight road every 12km—certainly a negligible aspect given the challenges to come.

45–70km: Steady degradation

Around 45km, my left thigh put itself into the center of attention. While not strongly hurting per se, it felt like a foreshadowing of future muscular fatigue and failure that seemed more appropriate after 37km in a marathon rather than the first half of an ultra. While I knew that losing time before the 60km-mark was dangerous, the split after 50km offered convenient mental relief: A time of 3 hours and 23 minutes meant a seven-minute buffer. Hence, when Christian joined after 50km, the goal was to stay around a 4:10 pace to sustain this buffer until the inevitable slow-down due to the increasingly fatiguing legs.

The last 30km: How do my legs still manage to go on?

Around 70km into the race, the constant twitches and muscular pain had evolved into a feeling of complete fatigue in both thighs. While the suffering itself at this point was not surprising (although I had hoped for 10 more somewhat pain-free kilometers), the degree of it made me skeptical. I certainly would have stopped in any training run, I even was event doubtful that I could keep going for more than 10km at any pace.

At this point, the mental race started: If the mind can endure, the body will somehow proceed along. While conveying my doubts about finishing to Janosch, he paraphrased Sabrina Stanley with something along the lines of “pain isn’t increasing linearly [with distance]—eventually, it will plateau”. This really helped—what better way to test this aphorism than to just keep going ;).

My muscles are as hard as week-old cafeteria bread. I can’t believe these are really my muscles.

Haruki Murakami
What I Talk About When I Talk About Running

The pain certainly passed all levels I would have deemed acceptable before the start. While Murakami’s image describing his legs after 75km of the Lake Saroma 100 is certainly memorable, it feels too passive and not intense enough to me. I remember it more like the feeling of a blood pressure cuff continuously expanding, but inducing small stitches instead of numbness at the thighs.

Looking on the bright side, the refueling was working smoothly6 and I was not alone on the road. Christian and Karsten were in front of me for the final lap, as well as Janosch and Theo on the bike behind me—urging me to keep going and not slow down.

The last ninety minutes were a constant cycle of slowing down, shouting with pain and forcing my legs to do a handful of “fast” steps to catch up, staying at a constant pace for a bit before succumbing to the pain and getting dropped again. Rinse and repeat—about every 200m. This certainly seemed unsustainable, like catching up with a cycling group that is drafting, before being dropped due to lack of power to stay in the wake.

Yet entering the final hour, somehow the motivation was rising. At this stage, it was OK to be suffering. Verbalizing the pain helped and, as hoped, it had leveled off rather than increasing by the kilometer. Being able to speed up, even for only 10 seconds, showed that there was still some gas in the tank. And although each kilometer felt longer than the entire first lap, the splits were still between 4:10 and 4:20. With around six minutes of buffer remaining, a sub-seven finish became increasingly plausible. On the 5km home stretch, a smoother pace and even a small acceleration to a 4:04-pace at the end was possible.

Tumbling across the (virtual) finish line

Finally, after 6:54:00 the 100km were over and in the books. Objectively, this was faster than expected before the race and certainly respectable for the first competitive ultramarathon. Subjectively, it certainly felt longer and entailed more suffering than my previous 10 marathon races combined, as well as shifted my perception of (sustainable) pain.

Regarding the balance—was it worth it? Overall, yes. Do I plan to do it again? (Probably) no, and almost certainly not faster.

The awesome support crew after the finish: Karsten, Christian, Olli, Theo and Janosch with a pretty destroyed version of myself in the middle.

A big shout-out to the crew: They were the enablers of this feat, and made it better than most races could have been. While, in the end, my legs needed to go the distance, finding and maintaining a mindset to push them forward was certainly easier with friends “forcing” me to.

One week onward, the feeling of my legs is still a reminiscence of the fight between mind and body over the last third of the race. Yet it also feels good to having successfully pushed a little closer to my personal limits as a recreational runner.7 And while I wonder if I ever pushed quite hard enough on my previous races, it seems good to test multiple points on the Pareto frontier ;).

  1. While a great experience, running the 60km-leg at the adidas Infinite Trails last year did not feel like pushing to the limit—not the least due to my reluctance to go all-out on a downhill. ↩︎

  2. But it would also be hard to assemble a representative dataset on the distributions of 100km finish times conditioned on marathon PRs. And marathon PRs might not be that predictive anyway. ↩︎

  3. The number of seven-hour finishes each year is certainly low enough to be skeptical. Additionally, the time of the elite national women have usually been a good match for mine over the marathon distance. Yet the German female record is 7:18, which made me even more doubtful of being 20 minutes faster. ↩︎

  4. Given the German weather and my preference for reasonably warm temperatures, a later date seemed to risky. ↩︎

  5. I had originally planned to do one 60km run, but the last 50km-run already suggested I had arrived at an overreaching state, so I opted for skipping it. ↩︎

  6. After the first hour, I had one Powerbar gel every 25 minutes. In the later stages, this was accompanied by some gel chips in the cheek to trick the body into expecting even more energy. ↩︎

  7. While I am proud of the achievement on a personal level, I want to make clear that I do not want to oversell this on an absolute performance level. Many runners have been faster (the German record is a 6:24), partly on harder courses . And as a comparison, many in the trail running community regularly accomplish much more strenous races in hot and humid conditions or ones that may last for days and are repeated within a couple of months↩︎

Henrich Kolkhorst
brains, robots, running

My research interests include brain–machine interfaces and human–robot interaction. Aside, I’m a competitive runner and triathlete.