#933 1984 · Barry Marshall (Royal Perth Hospital) · Gastroenterology / medical research
Marshall drank a dish of bacteria himself since no ethics board would test it on patients
the problem
Marshall believed a bacterium caused ulcers, but consensus blamed stress and acid, and no animal model proved it
background
By the early 1980s, gastroenterologist Barry Marshall and pathologist Robin Warren had found a spiral bacterium, later named Helicobacter pylori, in the stomachs of nearly every ulcer patient they biopsied, and they suspected it was the actual cause of a disease the medical establishment attributed to stress, spicy food, and excess stomach acid. But suspicion built on correlation wasn't proof, and every standard way to demonstrate causation was closed to them: the bacterium wouldn't infect guinea pigs, rabbits, mice, or piglets in the lab, so there was no animal model to test it on, and without one there was no path to a funded, ethically approved human trial either — regulators and grant committees weren't going to authorize infecting a patient with a bacterium to see if it gave them ulcers.
Waiting for an animal model that might never work, or for a research establishment that didn't believe the hypothesis to fund human testing anyway, could have delayed proof for years, if it arrived at all, while ulcer patients kept being treated for stress instead of an infection. Marshall realized there was exactly one subject whose informed consent didn't require anyone else's approval: himself.
what everyone would do
Keep searching for an animal model that would replicate the infection, or keep lobbying funding bodies and ethics boards for approval to test the hypothesis on human volunteers — both routes gated by other people's willingness to believe a theory the field considered already settled.
what they saw
Marshall saw the obstacle was procedural, not scientific: no institution would consent on a stranger's behalf to an unproven risk. He didn't need permission if he was both experimenter and subject.
the move
In 1984, Marshall drank a broth culture containing roughly a billion H. pylori bacteria taken from a patient's biopsy. Within days he developed nausea, bad breath, and vomiting, and an endoscopy at day ten showed his previously healthy stomach lining was now infected and inflamed, fulfilling Koch's postulates for infectious causation on a single, undeniable data point: himself.
why it works
The self-experiment works because it sidesteps the actual bottleneck, which wasn't scientific uncertainty but institutional risk-aversion: no committee has to approve exposing a subject to an unproven risk when the subject and the experimenter asking for consent are the same person, and informed consent has to come from the subject, not an institution vouching for a stranger. Because the result was unambiguous and dramatic on one well-documented body, it generated the kind of undeniable case study that reopens a settled scientific question faster than another decade of correlational data could have.
the payoff
The self-experiment produced undeniable proof; Marshall and Warren won the 2005 Nobel Prize in Medicine.
where it breaks
It only produces a single data point, with all the statistical weakness that implies, and it works only for risks the experimenter can survive and is willing to accept — it offers no template for testing anything with a realistic chance of permanent harm or death. It also depends on the self-experimenter already having enough scientific standing that their word and documented results will be believed once the stunt is over; without that credibility, the same act reads as reckless rather than as proof.
what came after
Ulcers, once treated with lifelong acid-suppressing drugs and dietary restriction, are now routinely cured with a short course of antibiotics, a shift in standard of care directly traceable to Marshall's self-infection.
references
- [1]Gut FeelingLasker Foundation, 2005laskerfoundation.org