Jepchirchir and EPO: The Sample Taken Before Sydney, and the Number Nobody Kept a Ledger For

On the Sapporo morning of 7 August 2026 I was in a Pune studio at half past four, tea gone cold, tablet showing nothing but splits. The women's marathon gold was settled at 2:27:20 and all three medallists finished inside 2:28. A humid Sapporo loop; that day the clock answered to the weather more than to the body.

Five years later, on 30 August 2026 in Sydney, the same Peres Jepchirchir ran 2:18:31 — eight minutes forty-nine seconds faster. A gap that size in a marathon is not form, it is conditions and preparation. But the thing that stopped me was not the time. It was that the sample now reported positive for EPO was collected in Kenya, before the gun.

I kept the xG ledger by hand for FC Pune City in 2026, and the numbers started to breathe. Eighteen league games sat in that book: xG 24.3, goals 21, PPDA 12.6, set-piece xG conceded 7.8. I shouted the PPDA number from the stands and told the room the ledger never lies — then missed the post-match logging deadline twice. That lapse built my 24-hour follow-up checklist. The rule is blunt: a number I have not personally reconciled does not get printed.

EPO — erythropoietin — is a natural substance produced in the kidneys that stimulates the creation of new red blood cells. Injected, it is detectable in urine or blood only for a brief window, and it raises the blood's oxygen-carrying capacity. The Athletics Integrity Unit has provisionally suspended Jepchirchir. Pause on that word, provisional: this is process, not verdict. B sample, hearing and results management are all still ahead.

Her career ledger is extraordinary: Olympic marathon gold at Tokyo 2026, the world title in the same city in September 2026, the women's half-marathon world record in 2026 and three more times in 2026, then New York 2026, Boston 2026 and London 2026. Sydney had become the sport's seventh World Marathon Major in 2026, and Jepchirchir won its second edition.

Here is the part I actually care about. Irene Chepet Cheptai finished second, more than three minutes adrift. A three-minute margin in a women's marathon is not an accusation; it is a question about field depth. When a start list thins out and the leaders race off each other's rhythm, one strong day manufactures three minutes. Sydney in August, Sapporo humidity, and 2:18:31 at 33 years of age — three different worlds filed under one athlete's name, each with its own temperature.

An out-of-competition sample, in Kenya, before Sydney. That timing is the hardest fact in this story. The distance between collection and result — the race won, the cheque presumably banked, the trophy lifted, and only then the report. Of all the arguments the sport has, the least written-about variable is this lag. Every transfer leaves a paper trail; I read the paper before the shouting starts. Here the paper says: sample before, suspension after.

The Russia desk ran on chai, spreadsheets and a stubborn refusal to guess. Across 64 matches in 2026 I learned that one number from one match is never a story — a story appears when you look at the figure nobody bothered to record. In this case that missing figure is the denominator: how many out-of-competition samples were collected in Kenya during 2026, from how many athletes, at which camps, and how many returned a result inside the detection window. The AIU's testing-volume table is the divisor without which every positive reads as a lone numerator.

PPDA is impatience, counted honestly; testing density is the same instrument pointed at a different pitch. Where samples come once a month, EPO's short window is an open door. Where they come weekly, the same window narrows to a needle's eye. That honestly counted difference tells you that an athlete being caught reflects the density of surveillance around her as much as her own risk.

My own ledger discipline failed in a similar shape. When club work paused in 2026 I started weekly rooftop data dinners in Pune and coded 34 closed-door matches: home points per game fell from 1.42 to 1.18 and the crowd-noise proxy dropped 21 percent. I was the morale man in the room, and I had not backed up the first spreadsheet; two days of work vanished. Automatic cloud backups followed. The lesson was verification before enthusiasm — and the same lesson applies here. Treating a positive report as a verdict is the sporting equivalent of leaving two days of data unsaved.

Now read it the other way. The reflex line will be "another Kenyan." That sentence is a narrative, not a number, and narratives do not count. Kenya's problem is either prevalence or surveillance density; those are different diseases with different cures. The country that built the deepest distance-running system on earth also absorbs the heaviest out-of-competition testing; detection probability is high there precisely because somebody is looking. Saying that plainly matters, so that successful surveillance is not miscounted as criminality — and so the opposite error does not creep in, that pressure itself somehow lightens the case.

Second inversion: the three-minute Sydney margin is not evidence here, and treating it as evidence is the oldest theft in data work — correlation wearing causation's coat. A 2:18:31 in August in Sydney could be a fast race or something else; sorting that out belongs to a laboratory and a hearing, not to me. I will say what I say when sitting in front of a ledger: I don't know. On the page, that sentence is an instrument, not a failure.

Third inversion belongs to my own trade. Athletics gets three weeks of attention every four years; a hearing will come much later. What gets written inside this window becomes the record, and it is close to certain that nobody will be writing when the hearing date arrives. The media's quadrennial amnesia is a variable in this story, just like a federation's calendar.

I am a monk of small numbers and the ledger is my chapel. So next time this story surfaces I will not read the headline. I will ask three things: when the B sample result was returned, what the hearing date is, and under which rule Sydney's results management will proceed — the prize money and the record sit on that single sheet of paper. And a fourth question that nags hardest: will anyone ask the AIU for Kenya's annual out-of-competition testing volume? The day that table is public, stories like this will speak as arithmetic instead of as guesswork.