Thirty-two kilometres into my first marathon, my legs stopped taking instructions. Not slowed down — stopped, the way a phone dies mid-sentence instead of fading out with a warning. One second I was holding a pace I’d trained for; the next I was walking past a water station I hadn’t planned to stop at, doing arithmetic on how many kilometres were left and not liking any of the answers. I had taken one gel that whole race, somewhere in the first hour, because it was in my pocket and I felt like I should probably do something with it at some point. That was the whole plan. I walked the last ten kilometres convinced I had a training problem, when what I actually had was a fuelling problem I had never once rehearsed.
The first one: nothing until it was too late
I had done the long runs. Eighteen, twenty, twenty-two kilometres on weekend mornings, feeling strong, telling myself the taper would carry the rest. What I had not done was eat anything meaningful on any of them — a training plan tells you how far and how fast, not what to put in your mouth while you do it, and I filled that gap with nothing and called it toughness. On race day, adrenaline and three days of half-hearted carb-loading bought me about two hours. After that the tank was simply empty, and wanting to run faster does nothing to an empty tank. The wall at 32K was not bad luck. It was the bill for twenty-two-kilometre long runs fuelled by willpower and a bottle of water.
The second one: overcorrecting into a different mistake
The obvious lesson from marathon one was “eat more,” so for marathon two I did — six gels stuffed into a belt instead of two, starting at 20 minutes and going roughly every 15 after that, on top of whatever sports drink was on the course. By 25K my stomach had had enough and made sure I knew it. I did not hit a wall that day so much as spend the last third of the race negotiating with my own gut, slowing to a pace that had nothing to do with my legs and everything to do with not throwing up in front of the mile-18 crowd. The mistake had changed shape but the cause underneath it was identical: I was still deciding in the moment. This time the in-the-moment decision was “more is safer” instead of “I’ll get to it later,” and neither one was a plan I had tested — both were guesses dressed up as strategy.
The third one: rehearsed everything except the last ten kilometres
Marathon three, I thought I had actually solved it. I practised a fuelling schedule on long runs — a gel every 20 minutes on the clock, water only, no course drinks — and it worked cleanly out to about 28 kilometres, which is as far as my longest training run had gone. Race day added fourteen more kilometres of territory I had never tested, on a warmer morning than any of my long runs had been, and somewhere past 32K, tired enough that the schedule stopped feeling urgent, I let one gel slide because I “felt fine.” I felt considerably less fine twenty minutes later, limping through the last 5K on a plan that had been good right up until the point it mattered most. The training-run distance had quietly become the ceiling on the plan, and nobody had told the marathon about the ceiling.
The pattern, once I stopped being annoyed about it
Three different failures, one shared root cause: every bad decision got made during the race, by a brain that was tired, oxygen-starved in the practical sense, and increasingly willing to bargain. “I’ll just take one gel later” and “I feel fine, I can skip this one” are the same delusion in different clothes — both are a promise your race-day brain makes to your thirty-kilometres-from-now brain, and the second one never actually agreed to it. A fuelling plan that depends on a real-time judgement call is not a plan. It is a hope with a schedule attached, and hope does not survive contact with thirty-two kilometres of accumulated fatigue and a marching band you’re too tired to enjoy.
Why the schedule has to beat the feeling
Muscle glycogen — the carbohydrate stored in the muscles themselves — is finite, and for most runners it is running low well before the marathon is over, which is the entire reason gels exist in the first place. The trouble is that the feeling of running low doesn’t arrive as a helpful early warning; it arrives as the wall itself, at which point a gel is playing catch-up against a deficit it cannot close mid-race. By the time “I should probably eat something” turns into an actual thought, the moment for that thought to be useful has already passed. That is not a willpower problem and I stopped treating it like one. It is a timing problem, and timing problems get solved by clocks, not by paying closer attention to how you feel — the whole reason a schedule works is that it doesn’t ask a depleted brain to notice its own depletion in real time.
What a boring, rehearsed fuelling plan actually looks like
The plan that finally worked was almost aggressively unglamorous. Work out the number in advance — carbohydrate per hour, times expected hours, divided by grams per gel — on the gel calculator, so the count exists on paper before the taper starts, not as a guess made somewhere around kilometre 20. Standard guidance sits at 30–60 g of carbohydrate an hour for most efforts, with well-trained runners able to push toward 90 g using gels that blend glucose and fructose; whichever number applies to you, it should come from your realistic finish time, not from whatever felt right during marathon two. Then set an actual alarm — watch, phone, whatever survives four hours of sweat — for every 20 to 25 minutes from the start, and take the gel on the alarm, not on the feeling. The entire point is removing the decision from a brain that will, by kilometre 30, cheerfully talk you out of it.
Carry at least one gel beyond the number the calculator gives you for your goal time. All three of my bad races went wrong in the final third, which is precisely when a slower-than-planned pace means the race is taking longer than the original maths assumed. Someone on pace for 3:55 who slips to 4:15 needs roughly two more gels than the plan they printed the night before, not zero, and the spare is what covers that gap without another mid-race negotiation. Round up, not down — a gel you didn’t need weighs nothing; a gel you didn’t carry costs you the last five kilometres.
Rehearse past where it previously broke, not up to where training happened to stop
Rehearse the actual plan — same gels, same spacing, same fluid, same brand if you can manage it — on a long run that goes past the distance where things previously went wrong for you, not merely up to whatever your training block happens to top out at. Marathon three failed at 32K because I had only ever tested fuelling out to 28K; the last ten kilometres nobody rehearses in training are precisely the ten kilometres a marathon spends deciding whether the plan actually holds. The gut absorbs carbohydrate at pace better with practice, not by accident — it is a trainable skill sitting underneath a nutrition plan, and skipping the rehearsal is asking an untrained system to perform under exactly the conditions it has never seen.
The rest of the ledger
Gels are only part of the total energy a marathon spends. Running the same finish time and body weight through the calories burned calculator puts a number on the full gap — for a lot of four-hour marathoners that lands somewhere around three thousand kilocalories, against a race-day gel plan that covers well under half of it. The rest gets paid for by whatever you arrived at the start line already carrying: the pre-race meals, the carb-loading in the two or three days before, the taper that actually let your legs store what you fed them. A perfect gel schedule cannot rescue a runner who skipped that part, which is its own honest asterisk on everything above — the alarm on your watch is only doing its job if there was something worth releasing when it goes off.
None of this is complicated, which is exactly why it took three marathons to take seriously. Work out the count before the taper starts, put an alarm on it instead of a feeling, rehearse it past the distance where it previously broke, and let the plan be boring enough that tired-you at 32K has nothing left to negotiate. Skoma turns the finished training block into a book once the counting — and the eating — is finally, finally done.