In 1747 a naval surgeon named James Lind was at sea with a crew suffering from scurvy, a disease that killed more sailors in that century than battle did. Nobody knew its cause. Everybody had a remedy, and all the remedies had supporters.
Lind did something unusual. He took twelve sick men whose cases were as similar as he could find, kept them in the same part of the ship on the same diet, and divided them into six pairs. Each pair received one of the six remedies, and nothing else was allowed to differ.
The pair given oranges and lemons recovered so quickly that one of them was back on duty within a week.
What made this work was not the fruit. It was the structure. Six remedies had been argued about for years by comparing them with nothing — each doctor remembering the patients who recovered under his own care. Lind compared them with each other, at the same time, under the same conditions.
One piece was still missing. Lind chose which pair got which remedy; he did not decide it by chance. That last idea — letting a coin assign the groups, so that even the differences nobody has thought of are spread evenly — arrived roughly two hundred years later, and it is what turns a careful comparison into a proof of cause.