What's in Bottled Water
Two studies six years apart reported counts a thousandfold apart, and the difference is mostly the microscope. What has been found in bottled water, and what a filter in a home kitchen is actually built to remove.
The question a reader actually asked
A reader wrote in with a short message: she had seen a headline about bottled water, had bought a filter jug the same afternoon, and wanted to know whether she had wasted the money. That is a better question than the headline was, and it turns out to have three separate answers depending on which part of it you take first.
So: what has been counted in bottled water, how the counting works, and what a filter in a home kitchen does and does not do about it. In that order, because the last one only makes sense after the first two.
What was counted, and by whom
The survey most people are thinking of ran in 2018 and was published in Frontiers in Chemistry — DOI 10.3389/fchem.2018.00407. The team tested a few hundred bottles across eleven brands bought in nine countries, tagged the contents with a dye that binds to polymer, and counted what lit up. Nearly all the bottles showed something. The averages ran to a few hundred particles per liter at the smaller sizes the dye picks up, and to roughly ten per liter in the larger size range, of which a second method confirmed about half as polymer.
Then in 2024 a group in PNAS — DOI 10.1073/pnas.2300582121 — pointed a much better instrument at three ordinary American bottled water brands. Their technique images single particles chemically rather than counting glowing dots, and it reaches down into a size range the older surveys could not see at all. Their figure was around a quarter of a million particles per liter, and about nine tenths of those were in the smallest range, the one nobody had been able to count before.
The gap between those two numbers is the single most useful thing on this page, and it is not what it looks like. Bottled water did not get a thousand times worse between 2018 and 2024. The microscope got better. Every count of this kind is a floor set by the instrument, and a rising number across the years is a story about detection before it is a story about contamination.
Where it comes from
Some of it is the bottle. Caps grind against threads, the polymer flexes, and a bottle that has sat warm in a car trunk or been opened and reclosed twenty times is not the bottle that left the plant. Some of it is the water before it ever reached the bottle. Some of it is the air in the room, which is why the careful studies run blank controls and subtract them, and why a study that does not run blanks is not worth reading.
Tap water has been surveyed too, less rigorously, and generally comes out lower per liter than bottled. I am not going to turn that into a recommendation, because the surveys are not comparable to each other and a lot of the tap water reporting was never peer-reviewed. What can be said is that nothing in the published record supports the idea that bottled is the cleaner of the two on this particular measure.
What a home filter is built to do
Here is the part that surprised me. The certification marks on a filter jug or a tap-mounted unit are written around specific things, and none of the common ones is a particle count.
- The everyday standard covers taste and odour — chlorine, mostly. That is what a basic jug is engineered for, and it does it well.
- A second standard covers a named list of contaminants such as lead and certain parasites. A filter carrying it has been tested against that list and nothing else.
- A third covers whole reverse osmosis systems, which is a different machine entirely.
So a jug filter with a certification on the box has not been tested against plastic fragments under that mark. It may well remove a lot of them; carbon block media catch particles by physical filtration whether or not anybody certified them for it. But "certified" and "tested for this" are not the same sentence, and the box is careful not to say the second one.
What the different filters plausibly do
Ranked by what the physics allows, not by what has been demonstrated in your kitchen.
Reverse osmosis pushes water through a membrane with openings far smaller than any of the particles discussed above. It should remove essentially all of them, and it is the only home setup where the mechanism makes the answer obvious. It also wastes water, needs plumbing, and strips minerals, which changes how water tastes and how it behaves in cooking.
Carbon block cartridges are solid, with a rated pore size, and will physically stop particles above that rating. The rating is usually printed. It is a real filter doing real filtration.
Loose granular carbon, which is what most cheap jugs contain, works by adsorption rather than sieving. Water finds paths between the granules. It was never designed to be a barrier to particles and it should not be expected to act as one.
And every one of these lives inside a plastic housing, with plastic tubing, and the reservoir of a filter jug is itself plastic that gets scrubbed, knocked and left in sunlight. Nobody has published what comes off the inside of a filter jug. That is a genuinely funny gap in the literature and it is not a rhetorical point — it is the obvious control experiment and it has not been run.
The kettle finding
One result from 2024 in Environmental Science & Technology Letters — DOI 10.1021/acs.estlett.4c00081 — is worth knowing because it involves equipment everybody already owns. Working with hard tap water, the authors boiled it and then filtered out the scale. Limescale forms as calcium carbonate crystals as the water heats, and the particles get caught up in the crystals and drop out with them. The reported reduction was large — most of what was there, in hard water.
The catches matter. It worked best in hard water, because soft water forms little scale and the mechanism needs the scale. It needs a filter step afterwards, even a paper coffee filter, because boiling only moves the particles into something you can remove. And it has been done once, in a laboratory, with tap water rather than bottled. It is a promising result, not a household instruction, and I am reporting it as the first rather than the second.
What none of this tells anybody
Everything above is a count. A count is a measurement of what is present, and it is silent on consequence. There is no published work establishing what quantity of plastic particles in drinking water produces any effect in a person, which means there is no threshold to be under, no target to hit, and no way to grade a filter as sufficient.
Anyone selling a filter on this subject is selling against an unspecified risk, and that is worth noticing before the card comes out. Removing particles you can measure is a defensible thing to want. Being told what it buys you is a different thing, and nobody is in a position to tell you.
What I do
Tap water, in a glass jug, with a carbon block cartridge on the kitchen tap that was installed for chlorine taste and stays for the same reason. Bottled water when traveling and not otherwise. The water in Smoked Paprika Lentil Soup comes out of the same tap as the water in everything else, and the soup has not changed.
The one habit I did change is storage: water sits in glass in the fridge instead of in a refilled bottle with a cap I had been grinding open twice a day for months. That costs nothing, and the reasoning behind it is at least mechanically sound, which is more than can be said for most of what gets recommended here.
What we still don't know
- How much of what is counted in a bottle comes from the bottle, as against the water before bottling. Nobody has separated the two cleanly.
- What any of these counts mean for a person, at any quantity, over any period.
- What home filters actually remove, measured by somebody independent, across the size range the newer instruments can see.
- What the filter housing and the jug contribute in the other direction.
Until those exist, the count is the only solid thing on the table, and a count is not a verdict. Claims Review: The Beet, Lemon and Ginger Morning Drink ended up in the same place from the opposite direction: there, people were claiming more than the evidence held; here, people are selling against a risk nobody has sized. Both are the same mistake wearing different clothes.
Editor's note: a report on drinking water and what has been published about it. Kitchen information, not medical advice, and no reason on its own to alter a prescription or a plan of care.