Municipal water disinfection (United States, 1908 to 1940)

Jersey City began routine chlorination of its drinking water in 1908 after a court fight, thousands of cities followed, and typhoid deaths fell steeply; economists still disagree about how much of the wider drop in deaths came from clean water as against sewers, milk rules, income and diet.

The record

Jersey City, 1908

• The Centers for Disease Control and Prevention (CDC) says Jersey City was the first U.S. city to begin "routine disinfection of community drinking water" in 1908, and that thousands of other cities and towns followed over the next decade. (CDC, History of Drinking Water Treatment; opened). The same CDC page also has a chart note dating the first continuous use of chlorine in U.S. water to "around 1915." The two dates are not reconciled on that page.
• A summary on Wikipedia describes the sequence. Jersey City sued the Jersey City Water Supply Company, a private firm, saying its water was not "pure and wholesome" as the contract required. Vice Chancellor Frederick W. Stevens ruled on May 1, 1908 that contaminated water reached the city two to three times a year, and his June 4, 1908 decree gave the company three months to propose a fix. John L. Leal, the company's physician and consultant, hired engineer George W. Fuller on June 19, 1908. Fuller built a chlorination plant at Boonton Reservoir in 99 days. It began operating on September 26, 1908, treating about 40 million gallons a day with chloride of lime. A second trial before Justice William J. Magie ran 38 days, and on May 9, 1910 he ruled for the company, finding chlorine an acceptable treatment. (Wikipedia, John L. Leal; opened. This is a secondary source; the court records were not opened.)
• The dose is given as 0.2 to 0.35 parts per million of calcium hypochlorite. Philadelphia's Belmont plant in 1913 was the first to use chlorine gas continuously. (Wikipedia, Water chlorination; opened)
• Spread: Wikipedia's Leal page says that by 1914 over 21 million Americans were served chlorinated water, and by 1918 more than 1,000 North American cities used chlorine, serving about 33 million people. Not checked against a primary source.

Earlier and parallel steps: filtration and sewage

• Lawrence, Massachusetts, began filtering river water in 1893 during a typhoid epidemic on the Merrimack River. Wikipedia says this led to a "marked reduction" in typhoid and in the overall death rate. (Wikipedia, Lawrence Experiment Station; opened. No numbers on that page.)
• Many cities filtered water first and chlorinated later, or the reverse. The table below shows dates for six of 13 cities studied by Cutler and Miller.

(Cutler and Miller, NBER Working Paper 10511; opened through a text summarizer, so the table should be rechecked against the paper)

• Cutler and Miller say modern sewage treatment was rare before the 1930s and 1940s and few of their 13 cities had it by 1936, so they could not measure it separately. Anderson, Charles and Rees say that nearly every major city already had sewers by the early 1900s, so sewers themselves could not be studied, but 15 of their 25 cities treated sewage by 1940. (NBER 10511; Anderson, Charles and Rees, NBER 25027; both opened)

Timeline

What happened to typhoid

The measures differ: CDC's figure is a national rate and may count cases; the two economics papers measure deaths in big cities. They are not comparable. Both papers agree filtration cut typhoid deaths a lot. Anderson and colleagues say Cutler and Miller's typhoid estimate is "nearly identical" to theirs.

What happened to deaths from all causes: the dispute

| Study | Sample | Main finding | Source | |---|---|---| | Cutler and Miller (Demography, 2005) | 13 cities, 1900 to 1936 | Clean water (filtration and chlorination) explains nearly half of the drop in total mortality, about three quarters of the drop in infant mortality and nearly two thirds of the drop in child mortality. Social return above 23 to 1; about $500 per life-year saved (2003 dollars). | PubMed abstract; NBER 10511 | | Ferrie and Troesken (Explorations in Economic History, 2008) | Chicago, 1850 to 1925 | Chicago's crude death rate fell 60%. Introduction of pure water accounts for roughly 30 to 50% of the decline; diphtheria antitoxin and milk inspection had much smaller effects. | NBER 11427 abstract | | Anderson, Charles and Rees (NBER 25027, 2018) | 25 cities, 1900 to 1940 | After correcting data and method, the filtration effect on total mortality is small and not significant (about 1 to 3%); on infant mortality about 11 to 13% lower; chlorination linked to an 8 to 10% increase in infant mortality, which the authors call counterintuitive; sewage treatment not significant. They say living conditions and nutrition are likely larger drivers, "an educated guess." | NBER 25027 | | Beach, Ferrie, Saavedra and Troesken (J. Economic History, 2016) | City data linked to 1900 and 1940 censuses | Eliminating early-life typhoid would have raised later earnings about 1% and added about one month of schooling. | NBER 20279 abstract |

(Cutler and Miller, PubMed 15782893; Ferrie and Troesken, NBER 11427; NBER 25027; NBER 20279; all opened at abstract level, plus the full text of 10511 and 25027 through a summarizer)

• Anderson, Charles and Rees list specific problems they say they found in the earlier study: clustering of standard errors, a Memphis 1916 gap coded as zeros, transcription errors in some mortality counts, and wrong chlorination and filtration dates for several cities. Cutler and Miller have posted a response, according to the NBER page; we did not read it.
• Cutler and Miller also report the Mills-Reincke effect: filtration was linked to fewer deaths from non-waterborne diseases such as pneumonia and tuberculosis, with a conservative multiplier of about 3 (NBER 10511, as summarized). The mechanism is debated; we did not find a source here that resolves it.

Cost

• Cutler and Miller estimate chlorination at roughly $150,000 to $200,000 a year in variable costs for a big city (from 1915 data, in 2003 dollars), with small fixed costs. Filtration is hard to isolate; they valued the mean big-city water system at just under $300 million (2003 dollars, 1915 data). (NBER 10511, as summarized)
• Who paid is not detailed in the sources we opened. In Jersey City the plant was built by the private water company under a court order. (Wikipedia, Leal)

Who was left out

• Werner Troesken's 2002 article asks how well American cities supplied Black residents with water and sewer service in 1880 to 1925 and what led governments to give any access, offering fear of epidemics and the degree of residential segregation as explanations. Our copy was the abstract only. (Troesken, Journal of Economic History, 2002, via IDEAS; opened)
• Cutler and Miller cite that work for the statement that minority neighborhoods often received piped water later than white neighborhoods. They say this would, if anything, make their estimates conservative. Mean Black population share in their sample was 10.6% in 1900 and 14.3% in 1940. (NBER 10511, as summarized)

Costs and harms of chlorination itself

• Chlorine reacts with natural organic matter to form byproducts, mainly trihalomethanes and haloacetic acids. The WHO has said their risks are "extremely small in comparison with the risks associated with inadequate disinfection." (Wikipedia, Water chlorination; opened)
• Science News reports that a 1974 EPA study found possible carcinogens in some tap water, many containing chlorine, and that this hastened the Safe Drinking Water Act. EPA now limits the byproducts, and long-term exposure to them has been linked to higher cancer risk. (Science News, "50 years ago"; opened)

What this case shows

1. A contract dispute became a public health rule. The Jersey City case began as a city suing a private water company over a contract clause. The court's findings were a public record, and the decision in 1910 let other cities adopt the method. The question "what is in our water" was answered in a courtroom, not by a household. (Wikipedia, Leal; court records not opened)
2. Ordinary people could see the result in death counts. Typhoid deaths fell from 47 to 4 per 100,000 in 13 big cities between 1900 and 1920, and filtration cut typhoid deaths by 36% to 46% in two separate studies. Public vital statistics made this checkable. (NBER 10511; NBER 25027)
3. The benefit reached small children most. Cutler and Miller attribute about three quarters of the fall in infant deaths and nearly two thirds of child deaths to clean water; Anderson and colleagues find a smaller infant-death effect (about 11 to 13%) from filtration. Even the lower figure falls on the youngest, who cannot vote. (PubMed 15782893; NBER 25027)
4. The effect lasted into adult life. Beach and colleagues find early-life typhoid exposure cost about 1% in later earnings and about a month of schooling. Clean water was an investment in people, not only an avoided funeral. (NBER 20279)
5. Not everyone got it at once. Black neighborhoods often got piped water later. Safe water came through city budgets and politics, which means access depended on who had a voice in them. (Troesken 2002 abstract; NBER 10511 citing it)
6. Fixes make new risks, which then need new rules. Chlorine saved lives and created byproducts, which in the 1970s led to federal limits. Both facts are public record. (Science News; Wikipedia, Water chlorination)

The strongest objections, and our answers

1. Objection: Sewers, milk rules, better food and rising income also cut typhoid and deaths, so clean water cannot take the credit. Anderson, Charles and Rees argue exactly this for total mortality and offer nutrition and living conditions as likelier drivers. The CDC page itself credits water treatment together with better source water and better sanitation and hygiene. Mitigation: we report both the 2005 and 2018 findings. On typhoid specifically, both teams agree the effect of filtration is large. For total mortality the estimates range from "nearly half" (Cutler and Miller) to 1 to 3% (Anderson et al.), and we do not pick one.
2. Objection: City comparisons are unreliable because cities adopted treatment at different times for different reasons. True. Cities that adopted early may have been richer or hit harder by epidemics. Cutler and Miller say their five-year-lead checks show no pre-trend, but Anderson and colleagues identify data errors in the earlier work. Mitigation: we show dates for six cities and say rates by city were available to us only as charts, not tables.
3. Objection: Chlorination was a minor step; filtration did the work. Anderson and colleagues find filtration, not chlorination, associated with lower infant deaths, and chlorination associated with higher. Cutler and Miller find the combination matters. Mitigation: we show both and note that cities often did both within a few years. Jersey City itself chlorinated without filtering, and no source we opened measures its typhoid rate before and after.
4. Objection: Typhoid was already falling before treatment. The CDC chart shows a long decline from 1900; a slide deck from a university course (not a primary or peer-reviewed source, so not relied on) argued the fall took 30 to 40 years and did not match a quick water effect. Mitigation: Cutler and Miller report a large drop in the year of adoption and effects that grew over five years; we did not see the city charts ourselves.
5. Objection: Jersey City is a famous story, but credit and risk are contested. Wikipedia's Leal article is a secondary source and the court records were not opened. Mitigation: we use it only for dates and say so. We did not verify claims about whether Leal sought a patent or about public reaction to chlorination, so we do not include them.
6. Objection: Chlorination by-products and today's pipes show the old fix has limits. Agreed. We include the WHO's statement that by-product risks are small relative to the risks of inadequate disinfection, and the 1974 EPA findings. We did not research present-day lead pipes or violations.

Open questions: what would change our minds

• Typhoid rates for individual cities before and after adoption were available to us only as charts. We have no city-by-city table. A full comparison needs the Turneaure and coauthors' Public Water-Supplies (1924) tables or city health department reports. Unverified.
• The dates in the six-city table come from Cutler and Miller's paper read through a summarizer. Recheck against the paper.
• Jersey City court details (the Stevens and Magie rulings, dates, 40 million gallons a day) come from Wikipedia. The court records and Fuller's and Leal's own reports were not opened. Unverified.
• The CDC page gives the Jersey City date as 1908 and also a chart note of "around 1915" for first continuous use. Not reconciled.
• Whether Jersey City's typhoid deaths fell after September 1908, and by how much, was not found.
• The claim of 21 million served by 1914 and 33 million by 1918 is unverified against a primary source.
• Cutler and Miller's response to Anderson, Charles and Rees (posted by the authors) was not read. The final peer-reviewed form of the 2018 paper was not checked.
• Troesken 2002's findings (as against its question) were not read; the full text was blocked. The extent of unequal access is therefore not quantified here.
• The 1914 Public Health Service bacterial standard and the early state rules were not researched.
• Who paid for early plants (rates, bonds, private firms) and what ordinary voters were told or voted on was not found in the sources we opened.
• Evidence that would change our minds: a clean city-level dataset with corrected dates that shows consistent drops in typhoid and all-cause deaths at adoption, or an analysis that shows sewers and milk rules fully explain them.

Sources

Found an error or a better source? We keep every correction on the record with its date and reason. Nothing here is quietly rewritten.