The chemicals making your e-cigarette taste like cotton candy may be harmful to your health, even when they're advertised as being safe. In a new study on the flavorings that go into e-cig liquids, researchers found that certain flavoring chemicals were present in quantities beyond what may be safe for inhalation. The study comes from Portland State University and is being published today in BMJ's Tobacco Control.

The researchers looked at 30 e-cig liquids — the fluid that's heated into vapor — taken from a number of companies, including the major manufacturers Njoy and Blu. All of the liquids were flavored, with tested varieties including tobacco, menthol, cotton candy, bubble gum, chocolate, and fruit flavors. A number of chemicals were present in multiple flavorings, including two that the researchers say were present in high enough volumes to potentially cause harm.

"A lot of e-cigarette manufacturers are representing that the components are food grade and implying that these are safe for use in e-cigarettes," Jim Pankow, a Portland State University chemistry professor and corresponding author of the study, tells The Verge. "But all of the safety studies" that they point to, he says, "have been for ingestion, not for inhalation." Neither Njoy or Blu assert that their liquids are "generally recognized as safe," but you'll quickly stumble across sellers who do if you're searching online for flavored liquids.

The trouble is, there isn't a lot of data on what is safe for inhalation. Pankow and his team honed in on two specific chemicals, vanillin and benzaldehyde, that recurred in the liquids and do have prior standards on inhalation safety for people working with them in potentially large quantities. In both cases, the researchers calculated that e-cigs may deliver double the exposure limit over the course of a day. That could be a problem, as both chemicals are known irritants of the respiratory tract, which enables breathing.

But ultimately, we still don't know exactly where the limit is before these chemicals cause harm. We don't know if someone would typically cross those limits while vaping, either: in these two cases, the calculated exposure is based on daily vaping projections, not measurements. The study also uses chemical measurements taken from the liquids — not from the vapor that they produce, which could make a difference.

"Several studies show that what is in the vapor can be different than what is in the liquid, ... and the vapor is what we primarily care about," Thomas Sussan, an e-cigarette researcher at Johns Hopkins Bloomberg School of Public Health who was not involved with the study, writes in an email to The Verge.

Pankow says that this is a fair question, but that there's no reason to believe the quantities will differ dramatically. "It's gonna be pretty similar. There's no reason to think that there would be a lot of destruction of the vanillin," he says. "So to a first approximation, just looking at the liquid is going to be a good estimate of what the exposure is."

Because these chemicals haven't been analyzed directly against human subjects or cells, Sussan cautions that "any implication regarding health is speculative and is based on projections of doses that may be achieved in a person." For now, he agrees, "the concentrations that are required to cause harm are largely unknown."

Still, there's at least one very obvious takeaway. "Maybe there should be labelling requirements for e-cigarette fluids," Pankow says, "that doesn't seem like a stretch."