London · July 1854
Two companies sell water in south London. The Southwark and Vauxhall Company pumps it from the Thames inside the town, where the sewers run in and the tide brings salt up from the sea. The Lambeth Company pumps it upriver, above all that. Their pipes run down the same streets, so one house drinks one water and the house next door the other, and nobody can see which.
Cholera came in July. Where someone has died, black cloth hangs on the door: crape. Most of it is inland, away from the river. The parish doctor says: the inland air.
You have a glass and a bottle of silver drops. Salt water turns milky with them; the town’s water is salt, the upriver water is not. Each glass costs you a house call. Where do you test, so that the doctor has to concede, and how few glasses does it take?
What your hands will do
Tap a house.
Add the silver drops.
Read the map.
Make the water stand alone: test houses so that the doctor cannot say “that is just the inland air”, with as few glasses as you can.
Make the water stand alone: test houses so that the doctor cannot say “that is just the inland air”, with as few glasses as you can.
Six streets, from the Thames up to the higher ground inland. The two by the river are crowded and poor; the two inland are thinly peopled and genteel.
The Board of Health asks what did it. Tap the sentence you will send; your tests go with it, and the Board reads the tests, not the sentence.
Make the water stand alone: test houses so that the doctor cannot say “that is just the inland air”, with as few glasses as you can.
Six streets, six houses each, from the river at the bottom to the inland side of town at the top. Where someone has died of cholera, black cloth hangs over the door: crape. Every house is on one of two pipes, and you cannot see which until you test it. Battersea water is the Southwark and Vauxhall Company’s, pumped from the river inside the town. Thames Ditton water is the Lambeth Company’s, from above the town.
Every tested house goes in: top row the houses with crape, bottom row the houses without; brown on the left, blue on the right. A house is hollow while its street has only one water tested, because the doctor can wave it away (“that street is inland”), and fills in solid the moment its street has both.
He speaks after every test. He will say that your brown houses are inland and inland is where the crape is, as long as that is true of your tests. He concedes when, in one street, you have tested a brown house with crape and a blue house, and the crape is at the brown. Your count of glasses is fixed at that moment; you may keep testing to see more of the map.
Whenever you like after the first glass, and at ten at the latest, write the Board one sentence: the inland air, the town’s water, or that these tests cannot say. The verdict reads your tests first: did the water stand alone? Then your sentence, judged against your tests and not against the answer. Then how many glasses it took.
The paper. John Snow (1856), Cholera and the Water Supply in the South Districts of London in 1854, reprinted from the Journal of Public Health, October 1856 (London: T. Richards). No DOI; the text is on Project Gutenberg as eBook 66507. Game built from this paper; the paper is the authority where the two disagree.
Between the epidemics of 1849 and 1854 the Lambeth Company moved its intake from the middle of London to Thames Ditton, above the tide and the sewers; the Southwark and Vauxhall Company stayed at Battersea Fields. Their pipes ran down the same streets, into the same courts, often to neighbouring houses, and supplied rich and poor alike: three hundred thousand people divided by nothing but their water. Snow went to the house of every cholera death in the mixed subdistricts for seven weeks, asked for the water-rate receipt, and where that failed tested the water for salt (the tide had made Southwark and Vauxhall water brackish, forty grains a gallon against one). In the first four weeks the Southwark and Vauxhall houses died at 107 per 100,000 against 8: thirteen to fourteen times. Over his seven weeks, 507 deaths in 16,038 houses against 98 in 20,554, about seven to one, and “nearly the same throughout” the subdistricts. The crowded, dirty, very poor riverside subdistricts of Lambeth Church and Waterloo Road, mostly on Lambeth water, died little; the thinly peopled, comparatively genteel Clapham and Battersea, on Southwark water, died much. The parish and the air had nothing to do with it once the water was known.
A cause that varies from house to house can be told apart from everything that varies from district to district, if you compare neighbours. Between the inland side of town and the riverside, a hundred things differ, and the water is one of them; any difference in deaths can be read as any of the hundred. Between two houses on the same street, one on each pipe, only the water differs. That is why Snow walked the mixed streets and not the all-Southwark or all-Lambeth districts, and why the Board of Health’s comparison of districts could be answered and his could not.
Mechanism kind: reconstructed. The map’s six streets are six of Snow’s subdistricts, ordered from the river inland: Waterloo Road (first part), Lambeth Church (first part), Trinity, Borough Road, Kennington (first part) and Clapham. Each has six houses, with the subdistrict’s real share of houses on the Southwark and Vauxhall Company’s water (23, 16, 55, 57, 85 and 98 per cent, Tables I and II), placed at random within the street so that a house is often on the other pipe from its neighbour, as the paper says. Each house wears crape with the paper’s probability for its water, deaths over houses supplied in the first seven weeks over the twenty-one mixed subdistricts (Table I totals), multiplied by ten so that a map of thirty-six shows it. The ratio between the two waters is the paper’s, about seven.
What the game demonstrates: that tests placed on one side of town cannot tell the water from the side, and tests placed in the same street can; and what the map looks like when the water is the cause and the pipes are sorted by district: crape inland, where the town’s water is, and in the middle streets at the brown doors only. What it assumes: everything a map of thirty-six has to invent: which house is on which pipe, which house dies (drawn independently at the paper’s rate), the same rate for every street (the paper reports it as nearly the same throughout), the floors on the draw (five to ten crape doors; more inland than by the river, which the tables’ shares give in six maps in seven unaided; at least one street with a brown crape house and a blue house without), the cap of ten glasses, the doctor’s rule of concession (one street with a tested brown house wearing crape and a tested blue house, the brown houses worse by the paper’s gap or no blue crape at all), the parish doctor and his inland air, and the sentence. The verdict judges identification by the tests alone: a right sentence without a same-street comparison is called a guess. The line under the map is deaths over tested houses by colour, as Snow divided deaths by houses supplied, once over all tests and once over the streets where both waters were tested. What the paper establishes: the counts and rates in its tables, and the identification argument. It does not establish anything about any particular house or street beyond the shares and rates it reports.
Built by Steffen Huck with Claude under the paper-to-game skill, November 2026; fourth build. Every number the game uses sits in a commented config block at the top of the source, tagged with its table or page. Type: Abril Fatface and Libre Caslon Text (SIL Open Font License), embedded.