What researchers actually found inside 100 South Australian rainwater tanks

If rainwater reaches your kettle or saucepan, its journey matters as much as its apparently clear appearance. A roof, gutter, tank and pipe can each contribute material too small to see—and different tests reveal profoundly different things about it.

Researchers sampled 100 domestic rainwater tanks across Adelaide and the Adelaide Hills between May 2024 and January 2026. Surveys were completed by 97 households; 67% of those respondents reported using tank water for drinking and cooking. The distinction matters: the water measurements describe 100 tanks, while the usage figure describes the people who answered the survey.

The useful question is not simply, “Was something found?” It is: what did the test actually establish?

The bacterial results were genetic signals

According to Adelaide University’s account of the study, DNA associated with Campylobacter was detected in 13% of tanks, Pseudomonas aeruginosa in 12%, Salmonella in 10% and Shigella in 3%. Bacterial DNA was also detected in some tanks supplying drinking water.

These are meaningful hazard detections, but they are not diagnoses of dangerous water or evidence that anyone became ill.

A genetic test can recognize a characteristic fragment even when no intact bacterium remains. Think of identifying a book from one torn sentence: the words may reveal which book they came from without showing that a complete, readable copy is present.

That creates an evidence ladder:

1. DNA detection: recognizable genetic material associated with a bacterium is present.
2. Viability: intact organisms are alive and capable of reproducing under relevant conditions.
3. Exposure: people encounter enough viable organisms through a plausible route.
4. Health outcome: the exposure is shown to have caused illness.

The new results reached the first rung in some tanks. The researchers said further work would be needed to determine whether the detected bacteria were alive and capable of causing infection. The study did not demonstrate an infectious dose, trace a bacterium’s route through an individual system or measure resulting illness.

Culture asks a different question

An earlier microbiological study of 53 Adelaide-region tanks used culture-based methods and detected E. coli in 28 tanks. Culture supplies evidence that organisms can grow under the laboratory conditions chosen for the test. DNA testing can detect genetic material without establishing such growth.

Neither method supplies a universal verdict on everything in the water. A culture result depends on which organisms the conditions allow to grow; a genetic assay depends on the sequences it is designed to recognize. The two Adelaide studies therefore cannot be treated as repeat measurements of the same thing.

The earlier field study also detected E. coli after filtration in some samples, although a filter tested under controlled laboratory conditions removed detectable bacteria. Its authors suggested several possible explanations—including installation, maintenance, exhausted cartridges or contamination after filtration—but did not establish which explanation applied to any particular household.

A hazard signal is not a record of illness

A double-blinded randomized trial helps show why the final rung of the ladder requires a different kind of study. In the Adelaide and Mount Barker trial, households already accustomed to drinking untreated rainwater received either an active treatment unit or a visually identical sham unit.

The rate of the trial’s defined gastrointestinal illness did not differ significantly between the groups. The active-versus-sham rate ratio was 1.05, with a 95% confidence interval from 0.82 to 1.33.

That result does not certify rainwater generally—or any particular tank—as safe. The trial examined established rainwater-drinking households and excluded people who were immunocompromised, had chronic diarrheal illness or used long-term antibiotics. It also predates the new sampling by many years. Its narrower lesson is methodological: finding a potential microbial hazard and showing that it causes additional illness are separate scientific tasks.

Metals are measured on another scale entirely

The new study also found a small number of tanks with cadmium, lead, arsenic or manganese above Australian health-based drinking-water guidelines. The available university summary does not provide enough detail to calculate how common each exceedance was or to extend those frequencies beyond the sampled tanks.

A metal result is not interpreted like bacterial DNA. It measures the concentration of a chemical element; there is no question of whether that element is alive. Its significance depends on the element, concentration, applicable guideline and water use.

Metal tanks were more likely than plastic tanks to have elevated metal levels, while systems reported as less well maintained were associated with higher concentrations. These are observational associations. They do not prove where a metal originated or reconstruct its route into a particular sample.

Solar panels offer a particularly instructive statistical wrinkle. The university summary says an initial association with elevated metals disappeared after researchers accounted for tank type. The primary study’s adjusted analysis instead reported a reversal: solar panels were associated with lower odds of elevated metals after adjustment for tank material, with an adjusted odds ratio of 0.265 and a p value of 0.026.

That result remains observational. It does not show that panels protect water, reveal a protective mechanism or justify adding panels as a water-quality measure. It says only that, within this dataset and model, tanks associated with panels had lower adjusted odds of elevated metals. The university summary’s claim that the relationship simply disappeared misses that reported direction.

Repeated samples reveal variation

A 2018 study of metals in Adelaide rainwater collected 365 samples from 53 tanks over 14 months. Forty-seven tanks produced at least one sample with lead above the cited drinking-water guideline, and roof material was associated with measured concentrations of lead, zinc, copper and cadmium.

That repeated design captured changes among tanks and sampling occasions. It cannot be placed beside the newer 100-tank study as a simple before-and-after comparison: the projects sampled different systems, at different times, using different designs and reported different sets of measurements. Together they show that tank-water chemistry can vary. They do not establish a regional improvement or deterioration over time.

Map the system without pretending to test it

A five-box sketch can turn the vague label “rainwater” into a documented collection system:

roof → gutter → tank → plumbing or filter → tap

For each box, record only observable facts: material, visible debris, the water’s route, the last known maintenance date, and any filter’s stated purpose and replacement schedule. Mark missing information as unknown.

This is reproducible documentation, not a home water-quality test. It cannot detect bacterial DNA, viable organisms or dissolved metals. Clear water is not a negative laboratory result, and a tidy gutter does not demonstrate safe drinking water.

The central finding inside these 100 tanks is therefore more precise than “rainwater contains contaminants.” Researchers detected particular genetic signals and measured particular chemical concentrations. Whether organisms were alive, how contaminants reached individual tanks and whether the water caused illness remain different questions requiring different evidence.