The Science of Marathon Records: How Athletes Keep Breaking the Impossible

When Elias Kipkorir crossed the finish line at the Rotterdam Marathon in April in a time of 1 hour 58 minutes and 42 seconds, he did not merely break the world record. He obliterated a barrier that physiologists had spent decades arguing was unbreakable. A sub-two-hour marathon in open competition — no pace car, no rotating phalanx of pacers, no sheltered course — was supposed to be impossible. Kipkorir made it look almost routine, running the second half faster than the first and crossing the line with enough left to raise both arms.

The time sent shockwaves through the sport, but it did not come from nowhere. Behind Kipkorir's performance lies a convergence of technological, physiological, and strategic advances that have been compressing marathon times for a decade — and that show no sign of stopping. Understanding how and why the record keeps falling requires a tour through the science of human endurance, from the carbon in the shoes to the oxygen in the blood.

Start with the shoes. The carbon-plate revolution, which began in earnest around 2017, has transformed marathon racing more profoundly than any single innovation in the sport's history. Modern racing shoes — stiff carbon-fibre plates sandwiched between layers of ultra-responsive foam — return energy to the runner with each stride, reducing the metabolic cost of running at a given speed by an estimated 4% to 6%. In a race decided by margins of seconds, a 4% efficiency gain is enormous. "The shoes did not make slow runners fast," said Dr. Ingrid Thorvaldsen, a biomechanics researcher at the Norwegian School of Sport Sciences. "But they made fast runners significantly faster. At the elite level, that is the difference between a great time and a record."

The shoes are only the most visible element. Behind the scenes, altitude training protocols have been refined to a degree that would have been unrecognisable a generation ago. Kipkorir and most of his elite Kenyan and Ethiopian contemporaries train at altitudes above 2,400 metres, where the thin air stimulates the body to produce more red blood cells, boosting oxygen-carrying capacity when they descend to race at sea level. The physiological response has been understood for decades, but the precision with which modern coaches manipulate the timing, duration, and altitude of training blocks — often using portable hypoxic tents to simulate altitude even when athletes are at home — has turned a blunt instrument into a scalpel.

Pacing technology has undergone an equally radical transformation. The days of the lone runner guessing their pace from kilometre markers are over. Today's elite marathoners race with GPS-enabled watches that provide real-time pace, cadence, and heart-rate data, and their coaches monitor biometric telemetry from the sidelines. In Rotterdam, Kipkorir's team used a proprietary pacing model — built on machine-learning analysis of his training data — that prescribed a negative-split strategy: start conservatively, bank energy, and unleash a devastating final 12 kilometres. The model predicted a finishing time of 1:58:38. The actual time was four seconds slower. "That is not coaching by feel," said Kipkorir's manager, Stephen Okello. "That is engineering."

Nutrition science has contributed its own marginal gains. The current consensus among endurance physiologists favours aggressive carbohydrate intake during the race itself — far more than was recommended even a decade ago. Kipkorir consumed roughly 120 grams of carbohydrates per hour during his record run, delivered through a combination of gels and a custom drink mix designed to maximise absorption without gastrointestinal distress. The approach, based on research showing that the gut can be trained to absorb higher volumes of fuel, keeps blood sugar elevated and delays the onset of the dreaded "wall" that has ended so many marathon dreams after 30 kilometres.

The cumulative effect of these advances raises the question that has fascinated exercise physiologists for a century: is there an ultimate limit to human marathon performance, and if so, where does it lie? The answer depends on whom you ask. Dr. Michael Joyner, the Mayo Clinic physiologist whose 1991 paper first modelled the theoretical fastest marathon, originally placed the limit at around 1:57:58 — a figure he has since revised downward as the shoes and nutrition science have changed the inputs to his model. Others are less certain. "Every time we define a limit, someone breaks it," said Thorvaldsen. "That does not mean there is no limit. It means we have been consistently wrong about where it is."

The physiological constraints are real. The human heart can pump only so much blood. Muscles can metabolise only so much oxygen. Tendons and ligaments can absorb only so much impact before they fail. At some point, the returns from better shoes and smarter pacing will diminish, and the hard ceiling of human biology will assert itself. But that point, wherever it lies, appears to be lower than anyone thought a decade ago.

Kipkorir, for his part, is already looking ahead. In a press conference after Rotterdam, he was asked whether the record could go lower still. He paused, smiled, and offered a single word: "Yes." His next race is the Chicago Marathon in October, where the flat course and autumn weather will provide ideal conditions. His training camp in Iten, Kenya, is already underway. Somewhere in the thin highland air, the next impossible time is taking shape.