The Tax Cut That Accidentally Lowered England's Death Rate
A 1784 law meant only to defeat tea smugglers pushed boiled water into English households at scale, and the parish records show a real, measurable drop in mortality that nobody who wrote the law intended or noticed.
In 1784 the tax on tea entering England stood at 119 percent, and the Treasury had very little to show for it. A rate that high does not raise revenue so much as redirect trade: more tea reached English cups through smugglers than through legal import, and the merchants who paid the duty were losing, at national scale, to organised competitors who did not. William Pitt the Younger, advised by Richard Twining of the tea family, put through the Commutation Act of 17842, which cut the duty from 119 percent to 12.5 percent in a single stroke, effective the following year. The reasoning was ordinary revenue arithmetic. A small rate on a large legal volume collects more than a punitive rate on a volume that has gone underground, provided the legal price falls far enough that the smuggler's margin disappears.
It did, immediately and by a wide margin. Legal tea imports went from 4,962,000 pounds in 1784 to 16,307,000 pounds in 17852, a jump that reflects smuggled tea shifting into the legal ledger rather than a genuine tripling of what England drank. Smugglers organised public opposition to the Act and lost: a trade that had supplied most of England's tea for a century could not survive a legal price it no longer undercut. On its own terms, the Act was a competent piece of fiscal policy, and for two centuries that was the whole of the story.
Why the leaf and not the bean
What made the price cut travel so far down the income scale was a property of the commodity rather than of the policy. Tea leaf can be reused. A household could brew the same leaf repeatedly, weaker each time, which decouples the cost of a cup from the cost of the purchase and therefore from the household's income in a way that few other goods of the period managed. Ale and beer had no such trick, and their cost had been pushed up by England's malt tax besides. Coffee, also traded by the East India Company, does not stretch in the same simple way.
The social geometry pointed the same direction. Coffee houses were male-only establishments. Tea could be bought and served by women, and the tea gardens were mixed-sex family venues, so the drink reached parts of the population that the coffee trade structurally could not. A cheap leaf that could be brewed thin, in the house, by anyone, becomes a universal drink rather than an elite one. That was not the Act's intention either, but it is why the Act's effects were general.
What the parish registers show
The consequence nobody legislated for is that tea, unlike the water it displaced, is made by boiling water first. Boiling kills the waterborne pathogens, dysentery and its relatives, then called the bloody flux, that were among the leading killers in England's crowded and rapidly urbanising towns. This was several decades before germ theory existed, so the mechanism operated entirely without anyone's understanding of it.
Historians had guessed at the link for years. The anthropologist Alan Macfarlane compared England with Japan in 1997 and concluded that tea "caused boiled water to be used, which caused dysentery to be minimized." The beer England was already drinking argued against the theory as much as for it: tea is less nutritionally useful than beer, so the decline of home-brewed beer under England's malt tax should have left the population worse fed, not better protected. Suggestive, but a hypothesis, and an unresolved one.
The economist Francisca Antman of the University of Colorado Boulder put numbers to it in "For Want of a Cup: The Rise of Tea in England and the Impact of Water Quality on Mortality,"1 published in The Review of Economics and Statistics, volume 105, issue 6, pages 1352 to 1365, in 2023, and circulated the year before as IZA Discussion Paper No. 15016. It is the first study to test the claim quantitatively.
The raw material is the parish record: burials, baptisms and marriages for 404 English parishes, compiled by Schofield and Wrigley in 2003, covering 1725 to 1834. Nobody measured water quality in the eighteenth century, so Antman proxies it with each parish's elevation and its number of nearby running water sources, then compares mortality trends in the worse-water parishes against the better-water ones, before and after 1785. A second strategy, running from 1761 (the first year national tea import volumes exist), interacts those import volumes with the same water-quality proxies.
Before 1785 the two groups of parishes show no difference in mortality trend, which is the parallel-trends condition the causal reading requires and the sort of check that frequently fails. After 1785 they diverge. By the end of the study period, mortality in the worst-water parishes had fallen by roughly 18 percent more than in the better-water parishes, and the gap widened over time as tea drinking spread through the population. The estimates are significant at the 1 percent level. Nationally, across the 1761 to 1834 window, tea imports rose from roughly 1 pound per person to almost 3 pounds per person, while England's crude death rate fell from about 28 to about 23 per 1,000, with the decline visibly accelerating after the tax cut.
The tests that could have failed
A correlation between rising tea and falling mortality across seven decades of the Industrial Revolution has an obvious rival explanation, which is that the period was improving in a hundred other ways at once. Antman's answer is a falsification test using London mortality by cause of death, taken from a contemporary compilation by Marshall in 1832. Rising tea imports track a statistically significant decline in deaths from waterborne disease, the flux and its kind. They show no significant relationship with deaths from tuberculosis, then called consumption, which is airborne and which boiling water cannot touch. A general improvement in the times would have moved both.
A second check splits deaths among London children by age, and the split is the point. Children aged two to five, old enough to have left the household's most protected care but still not meaningful tea drinkers, show no benefit from rising tea imports that the data can distinguish from none. Infants under two do show a decline, smaller than the effect on drinkers of tea themselves but real, which is what the mechanism predicts once boiled water is displacing contaminated water in the household generally rather than only in the cup a person drinks: an infant benefits indirectly from a cleaner kitchen even though nobody is pouring tea into it.
A public health measure nobody wrote
None of this was policy. The Commutation Act was a revenue and anti-smuggling measure, argued on revenue grounds, passed on revenue grounds, and vindicated on revenue grounds within a few years. Its authors were not concealing a health objective, because there was no health objective available to them. Germs would not be identified for another century, and the notion that a schedule of duties might function as a sanitation programme would have been unintelligible to everyone in the room.
Antman draws the modern lesson herself, and it is the load-bearing part of the finding rather than an ornament on it. The interventions that change a population's behaviour at scale are sometimes not health campaigns at all. They are changes that make a beneficial behaviour cheap and culturally desirable for reasons of its own, adopted voluntarily because the adopter wants the thing, not because a ministry has recommended it. A tax cut that made a boiled drink affordable to a labourer's household did more against waterborne disease than any instruction to boil water could have done, partly because nobody was being instructed.
The Act's own authors measured their success in the ledger of legal pounds against smuggled ones. The duty fell from 119 percent to 12.5 percent, the smuggling trade lost the price gap it lived on, and in the parishes with the worst water the burial rate fell about 18 percent further than in the parishes with the best.