In late August 1854, a few streets in London's Soho began to bury their neighbours. Cholera arrived suddenly and killed quickly. Within a few weeks roughly 600 people in the area had died. Nearly everyone with an opinion agreed on the cause: bad air, the "miasma" that rose from filth and rot.
One doctor, John Snow, thought the air was innocent. He suspected something in the water. He could not see the culprit, because nobody had yet shown that cholera was caused by a microbe. What he could do was draw a map.
Snow got the list of deaths and marked each one at the house where the person had lived, as a small black bar, stacking a new bar on the same house for each further death. One house ended up with 18 bars. When the bars were all down, they were not scattered evenly. They bunched around one spot: the water pump in Broad Street.
A cluster on a map is only a hint. The real work was in asking about the exceptions. Of 75 early deaths where Snow could find out, 68 were people who had drunk from the Broad Street pump, and only 6 were not. That is not proof of anything alone, but the stragglers were more telling than the crowd. A workhouse in nearby Poland Street had 535 inmates and only 5 died of cholera. It had its own well. A brewery close to the pump had more than 70 workmen, none with severe cholera. They had their own deep well, and were given malt liquor to drink. Meanwhile, people who lived far from the pump but sent for its water, because they liked the taste, fell ill.
That is the logic of the whole method: look for the people who should be sick but are not, and the people who should be safe but are not, and find what is different about them.
Below is an invented neighbourhood, not the real Soho map. Five pumps, a few hundred people, and one of the pumps is contaminated. Slide through the days and watch the cases appear. Then tap a pump to see how many of the cases live closest to it, and decide which handle to remove.
On 7 September Snow took his evidence to the parish board of guardians of St James's. The pump handle was taken off the next day, 8 September. It makes a tidy story: the handle comes off and the epidemic stops. The truth is messier. The outbreak had already been fading, as many residents had fled the area, so the lack of cases afterwards does not prove much by itself. The map and the exceptions made the case, not the handle.
The authorities were not won over by it. The idea that cholera spread through water did not become the accepted view until later, once microscopes and bacteriology caught up. But Snow's method outlived the argument. He had shown how to use a map to find the source of an outbreak, and the same idea underlies disease tracking today.
If you only look at where the sick people are, you can be fooled by crowds, by busy streets, by where most people simply live. Snow's trick was to compare each group with those who differ in just one way. The brewery and workhouse were a kind of natural experiment: same neighbourhood, same air, different water. Cholera followed the water.
In the lab above, check the brewery and workhouse, and the ringed houses that appear far from the guilty pump but still fetch water from it. Those oddities are what separate a hunch from an argument.