Plain Water: E. coli

In a water laboratory the sample bottles arrive in the afternoon, in a cool box, each one taken from a tap or a reservoir or a river that morning by someone in a van, and the first thing done with each of them is the same everywhere on Earth. A hundred millilitres is poured through a filter fine enough to hold back bacteria, the filter is laid on a dish of jelly that feeds them, and the dish goes into an incubator at thirty seven degrees, which is the temperature of a body, and is left overnight. In the morning the technician counts the dots. The dots that have turned the colour the jelly is designed to turn are colonies of a bacterium that lives in the gut of every person and every cow, and if there is one of them on the dish from a tap, the phone rings at the water works.

This article is about E. coli: what it is and why a harmless gut bacterium is the water industry's alarm, how it is counted and why the count is a day old, what the limit of zero means, what happened in a Canadian town in 2000 when the alarm was switched off, and why the bacterium the lab looks for is a messenger for everything it cannot see.

A gut bacterium

Escherichia coli was described in 1885 by a Munich paediatrician, Theodor Escherich, who found it in the stools of healthy babies, and it has been found in the gut of every warm blooded animal looked at since, in numbers of a hundred billion per gram of faeces, where it lives on what the animal has eaten and does, in nearly every case, no harm at all. It is one of the best understood living things, the workhorse of molecular biology, and the reason it matters to water is that it hardly lives anywhere else. Outside a gut it survives for days to weeks in water and soil and does not grow, so that its presence in a sample means one thing: faeces reached this water recently. Faeces is the source of nearly every waterborne disease on this site, cholera, typhoid, dysentery, hepatitis A, the rotavirus that kills infants, the parasites that the turbidity article describes, and each of them is hard to culture and some are impossible, so the industry looks instead for the organism that always comes with them and is easy to grow. E. coli is the indicator. It is not usually the disease.

The test

The count takes a day because the bacteria have to be grown to be seen. The oldest method, still used in many laboratories, is the multiple tube one: portions of the sample are put into tubes of broth with a lactose sugar that coliform bacteria ferment to gas, incubated, and the pattern of tubes that gas is read against a statistical table to give a most probable number of organisms in a hundred millilitres, which is the test in the opener of this article. Membrane filtration, from the 1950s, is faster to read: the filter on its dish grows one colony per organism and the technician counts them. The newest methods put the sample in a bottle of reagent that turns yellow if coliforms are present and glows under ultraviolet light if E. coli is, after eighteen to twenty four hours, and can be done by someone without a laboratory, which is why they have spread to small utilities and to the field. All of them say what was in the water yesterday. A works that waited for the result before deciding whether to chlorinate would be a day behind every failure, which is why the chlorine goes in regardless, the disinfection article on this site explains, and the test is the check that it worked.

The limit for drinking water is zero: not detectable in a hundred millilitres, in every sample, under the World Health Organization's guideline and in nearly every national rule. It is one of the few absolute limits in the industry, and it is absolute because the number of organisms that can infect is small and the dose that a sample happens to catch is random. A single colony on the dish is a failure. The response depends on where the sample came from: a repeat sample, a flush of the main, a check of the chlorine residual, and if it is confirmed and the cause is not found, a boil water notice to the district, which the myth about boiling on this site describes from the kitchen's side.

Walkerton

Walkerton is a town of five thousand in the farm country of southern Ontario, and its water in 2000 came from three wells run by a public utility whose two senior operators were brothers who had held the jobs for decades. Well 5 was shallow, in fractured rock, a few hundred metres from a cattle farm, and the utility had known for years that it was vulnerable to surface water and had not fitted the continuous chlorine monitor the rules required. On 12 May 2000, after days of heavy rain, manure from the farm washed into the ground around the well and into it, carrying a strain of E. coli called O157:H7, which cattle carry harmlessly and which in people attacks the gut wall and, in the young and the old, the kidneys, along with Campylobacter, another cattle organism. The chlorine dose at the well was too low to deal with it, and the operators' records, the inquiry found, had been invented for years: residuals written down that were never measured, samples labelled from one place and taken at another. When the laboratory reported E. coli in the town's water on 17 May, the operator told the health unit the water was fine, and the health unit, with children already in hospital, issued the boil water notice on its own on the 21st. Seven people died, about two thousand three hundred fell ill in a town of five thousand, and a number of the children were left with permanent kidney damage.

The inquiry, led by a judge, took two years and wrote two reports, the second of which is the rulebook that Ontario and much of the English speaking world adopted: mandatory operator certification, continuous chlorine monitoring on vulnerable wells, source water protection plans that look at the farm as well as the pipe, laboratories required to report a failed sample directly to the health authority, and a chain of responsibility from the operator to the minister. The brothers were convicted. The lesson that the inquiry put first was that the indicator had worked; the E. coli was in the sample and the sample was reported, and the system failed in what happened next.

The beach

The same count closes beaches. Bathing waters in Europe are classed every year by their E. coli and enterococcus counts, with limits of a few hundred per hundred millilitres for an excellent inland water and lower on the coast, and a beach that fails is signed and, after enough failures, closed; in the United States the state health departments do the same with their own numbers. The source is nearly always rain. A storm washes the streets, the dog walks, the farmland and, as the storm overflows article on this site describes, the sewers into the river and the sea, and for a day or two after it the count at the beach rises by a hundredfold and falls again as the sun and the salt kill the bacteria. The sign on the beach the morning after a storm is the membrane filter's count from the day before, and the people who swam that morning were in yesterday's water.

The messenger

The last thing to say about E. coli is what the test cannot do. It finds faeces; it does not find everything faeces carries. Viruses survive longer in water than E. coli and pass through soil that stops it, so a well can be clear of E. coli and carry the virus that caused the outbreak, which has happened; the parasite Cryptosporidium survives chlorine that kills E. coli, so a treated water can pass the test and carry the parasite, which is what Milwaukee learned in 1993; and the test says nothing at all about the arsenic, the nitrate, the lead and the PFAS that the plain water articles on this site have covered, because none of them came from a gut. The industry has added indicators, the enterococci, the spores of a bacterium called Clostridium that survive like the parasites, and the viruses that infect bacteria, and it has added the turbidity and chlorine measurements that work in real time. The E. coli count remains the first line in every report, because of what it reliably means: a gut, recently, upstream.

What it teaches

E. coli is the plain water article about the measurement the whole industry is organised around. A bacterium that harms almost no one is grown overnight on a dish so that its presence can stand for the presence of everything that does, and the limit is zero because the meaning of one colony is unambiguous. The test is a day old, so the water is treated as if it will fail. Walkerton is what happens when the test is passed on paper and not in the well, and the rules written afterward are the ones a town's operator now works under. The count at the beach after a storm is the same test, read for the same reason, by people who have learned not to swim until the number comes down.

Zero in a hundred millilitres, a day to grow the answer, and seven deaths in a town where the answer was written down before the sample was taken.