Since 2009 the people of Dimock, Pennsylvania, have insisted that, as natural gas companies drilled into their hillsides, shaking and fracturing their ground, their water had become undrinkable. It turned a milky brown, with percolating bubbles of explosive methane gas. People said it made them sick.

Their stories — told first through an investigation into the safety of gas drilling by ProPublica — turned Dimock into an epicenter of what would evolve into a national debate about natural gas energy and the dangers of the process of “fracking,” or shattering layers of bedrock in order to release trapped natural gas.

But the last word about the quality of Dimock’s water came from assurances in a 2012 statement from the U.S. Environmental Protection Agency — the federal department charged with safeguarding the Americans’ drinking water. The agency declared that the water coming out of Dimock’s taps did not require emergency action, such as a federal cleanup. The agency’s stance was widely interpreted to mean the water was safe.

Now another federal agency charged with protecting public health has analyzed the same set of water samples, and determined that is not the case.

The finding, released May 24 from the Agency for Toxic Substances and Disease Registry, a part of the Centers for Disease Control and Prevention, warns that a list of contaminants the EPA had previously identified were indeed dangerous for people to consume. The report found that the wells of 27 Dimock homes contain, to varying degrees, high levels of lead, cadmium, arsenic, and copper sufficient to pose ahealth risk. It also warned of a mysterious compound called 4-chlorophenyl phenyl ether, a substance for which the agency could not even evaluate the risk, and noted that in earlier water samples non-natural pollutants including acetone, toluene and chloroform were detected . Those contaminants are known to be dangerous, but they registered at such low concentrations that their health effects could not easily be evaluated. The water in 17 homes also contained enough flammable gas so as to risk an explosion.

The EPA had asked the ATSDR to help evaluate the health risks of its water samples back in 2011. At the time, the ATSDR warned people not to drink their water, and promised a more complete evaluation. The report released last month is that complete evaluation — an assessment of water samples taken from 64 homes during a short period in 2012 when drilling in the surrounding community had come to a temporary halt.

The fact that the contaminants were detected in water had been shared with residents in 2012. But the qualitative assessment of whether those contaminants pose a danger is new.

The new conclusions appear to call into question the EPA’s assurances, and could well reignite a smoldering controversy over whether the agency had prematurely abandoned its research into the safety of fracking in Dimock and other sites across the country. ProPublica reported in 2012 that the agency had curtailed investigations it had begun into potential water contamination in Pennsylvania, Texas and Wyoming.

The May 24 findings about Dimock also lead to an obvious question: How can two federal agencies charged with safeguarding the country’s health and environment use the same water samples and reach such different conclusions?

A spokesperson for the EPA offered a seemingly cryptic explanation: EPA was testing whether the contaminants were “hazardous,” while the ATSDR was considering whether they were safe to drink. In a statement the EPA sent to ProPublica, it described the ATSDR report as “useful information” for Dimock residents. The spokesperson promised that the agency would “consider the findings.”

Of course, the EPA, in certain respects, had already considered the findings. In 2012 — after collecting and analyzing the same water samples used by the ATSDR — the agency detected the same assortment of contaminants. It just ultimately did not describe their presence as a significant health threat. The agency eventually ended it investigation into the groundwater in Dimock. Its detailed water test results were distributed confidentially to homeowners without any evaluation of their meaning, while a general report was issued publically downplaying the danger.

At the time, ProPublica, which had been leaked a portion of the sample results shared privately with Dimock landowners, wrote that the water contained significant amounts of metals and chemicals and there appeared to be a disconnect between the EPA’s conclusions and the chemical make-up of the water. “I’m sitting here looking at the values I have on my sheet — I’m over the thresholds — and yet they are telling me my water is drinkable,” Scott Ely, a Dimock resident, told ProPublica in March, 2012. “I’m confused.”

One possible explanation is the difference in the two mandates and authority of the two agencies. The EPA — in examining the pollution in Dimock — was acting under the environmental law known as the Comprehensive Environmental Response, Compensation, and Liability Act. In other words, it was trying to determine whether Dimock’s contamination qualified as a federal Superfund site requiring a federal cleanup.

Pennsylvania environment officials had already determined that a number of drilling violations by a company named Cabot Oil and Gas had likely contributed to some pollution and disturbance underground. It sanctioned Cabot, suspended their drilling permits in the area, and required them, for a time, to supply drinking water to Dimock homes. In 2012, Cabot settled with dozens of families who had sued the company for an undisclosed amount.

That contamination existed, and that it may have been caused by the drilling, was not in dispute when the EPA made its judgment call. “Spills and other releases have been documented… from these drilling activities,” noted a January 2012 EPA document. “There is reason to believe that a release of hazardous substances has occurred.”

The EPA’s January action memo went on to describe the contaminants and summarized that “a chronic health risk exists for most wells and that the situation supports a ‘Do Not Use the Water’ action.” Beside several of the most concerning substances detected, the agency wrote: “note that children reside at this location.”

Still, Superfund law sets several criteria beyond a health concern to qualify for a cleanup, including the possibility of alternative solutions to a cleanup. There is also an explicit loophole built into the Superfund law for petroleum and natural gas; neither are allowed to be defined as a “hazardous substance” requiring cleanup.

How those factors weighed on the EPA’s characterizations of, and decisions concerning, Dimock’s water is unclear. The agency has never explained its decision in detail. But just six months later, the EPA issued a carefully-worded statement declaring what appeared to be a reversal, saying it “has determined that there are not levels of contaminants present that would require additional action by the Agency.” It said that it would stop delivering clean emergency water to Dimock homes, and noted that in the homes with the worst contamination residents planned to install water treatment systems.

The EPA’s 2012 decision was widely interpreted to mean that the water supplies in Dimock had been found to be safe.

“EPA finds Pennsylvania well water safe after drilling,” a headline in the Los Angeles Times read.

In a response to ProPublica’s questions this week, an EPA spokesperson said the agency had completed a second round of testing and concluded that the worst water problems were confined to just four homes. Those homes either already had, or planned to install, water treatment systems that the EPA expected would reduce their exposure risk. The EPA emphasized that ultimately cleaning rural residential water is not its responsibility. “Private drinking water well owners are responsible for sampling and maintaining their wells and addressing the operational issues that can affect drinking water quality,” wrote Michael D’Andrea, the EPA’s director of communications for its Mid-Atlantic region, in an email.

The ATSDR, on the other hand, serves to advise residents with regards to their health. It examined the same water samples as the EPA, but assumed that the highest levels of contaminants detected were what people would be exposed to. And it simply advises on issues of public health.

The ATSDR doesn’t address the stickiest political questions that have bogged down the conversation about “fracking.” It won’t require a cleanup, or address whether or not the contamination was caused by the drilling activities or some other source.

It also has limits; having only examined water sampled during a short period, more than four years ago, when drilling was on pause.

Today in Pennsylvania, drilling activity is down, thanks to the collapse of oil and gas prices. Statewide drilling activity is about one-fifth of what it was in 2012, and there is not currently any drilling in the immediate Dimock area. Cabot Oil and Gas is still banned from drilling new wells in a nine-square mile area there. Asked for comment on the ATSDR report, a spokeswoman from Cabot wrote “It is imperative to review this study in full context and next to the separately concluded investigations by both the DEP [Pennsylvania’s Department of Environmental Protection] and the EPA which found the level of contaminants found do not possess a threat to human health or the environment.”

Little additional water testing has been done in the Dimock area of late. The ATSDR says conditions may well have changed, and it would need more data to determine whether the water residents there are drinking today will make them sick.

But what its report does do is begin to explicitly answer one of the most important and enduring questions that has arisen as gas drilling has raised water contamination suspicions across the country: “ATSDR found some of the chemicals in the private water wells at this site at levels high enough to affect health … pose a physical hazard… or make the water unsuitable for drinking.”