How to Read the Numbers

A liter of bottled water held about 325 plastic particles in a 2018 survey and about 240,000 in a 2024 study^1,2. That is roughly 740 times more (our calculation), and both numbers can be right, although the 2024 count is disputed (see below). The teams looked for different sizes with different instruments in different bottles.

A microplastic number makes sense only once you know what was counted, how small the method could see, and whether anyone measured anything. Five questions sort that out in under a minute.

One bottle, two answers

Mason and colleagues tested 259 bottles from 11 brands bought in 9 countries, staining particles with Nile Red down to 6.5 µm¹. They averaged 10.4 particles per liter above 100 µm and about 315 more between 6.5 and 100 µm. The stain also lights up some non-plastic material, and particles under 100 µm were not confirmed by infrared, so small counts may run high (level C)¹.

Qian and colleagues used stimulated Raman scattering microscopy on 3 brands, looking for 7 polymers down to about 100 nm². They reported about 240,000 particles per liter, roughly 90% of them nanoplastics. It is a method demonstration, not a market survey.

The count is disputed (level D). In a published letter, Materić notes that ultrapure Milli-Q water run through the whole procedure held about as much plastic as the bottled water, and that the clean filters used as blanks miss contamination from handling³. The authors reply that the Milli-Q water was itself contaminated, with a different polymer profile, and that their filter blanks, exposed to the same lab air, controlled the remaining sources⁴.

Bottled water: reported particles per liter and the smallest size each count includes

Data table

label	value	unit	lower_limit	source
Mason et al. 2018, particles above 100 µm (Nile Red)	10.4	particles/L	100 µm	S-074
Mason et al. 2018, all particles from 6.5 µm (Nile Red)	325	particles/L	6.5 µm	S-074
Qian et al. 2024, all particles from about 100 nm (SRS microscopy; count disputed)	240000	particles/L	≈100 nm	S-033

Five questions to ask any number

1. What size range? A count with a lower limit of 100 µm and one with 100 nm are different quantities. If the size range is missing, the number cannot be compared.
2. Count or mass? Particles per liter and micrograms per liter answer different questions and cannot be converted without assumptions.
3. Measured or estimated? A measurement comes from analyzing a sample. An estimate comes from a model, often built on a handful of measurements.
4. Where, when, how many samples? Three brands are not "bottled water"; one cruise is not "the ocean"⁵.
5. Compared with what? "240,000 particles" sounds large until you ask what earlier methods could see and what dose matters for health.

Reporting guidelines ask authors to state exactly these things: size limits, blanks, recovery and sample numbers⁶.

Labs still disagree. In a 2019 exercise, 12 experienced laboratories analyzed identical bottles of spiked seawater, and even with best practice they undercounted particles smaller than 1 mm by about 20%⁷. In 2026, laboratories in seven countries using the same polymer standards and protocols still differed substantially in polymer identification and recovery⁸. Method choice alone can move a number (level B).

Count versus mass

Mass grows with the cube of diameter. One spherical particle 1 mm across has the same mass as 10¹² (a trillion) spheres of 100 nm made of the same plastic (our calculation). For polystyrene, that one sphere weighs about 0.55 mg (our calculation, density 1.05 g/cm³). A sample of nanoplastics can have enormous counts and almost no mass.

The two metrics can even rank sources in opposite order. A 2026 study of airborne microplastics from 5 to 100 µm estimated that land emits about 23 times more particles than the ocean (our calculation from its totals). Yet the ocean emits more mass, about 4,000 against 500 metric tons a year (model estimates, level C)⁹. The site keeps counts and masses apart.

Same mass, different count: one 1 mm sphere versus 100 nm spheres

Data table

label	value	unit	mass_mg
one sphere, diameter 1 mm	1	particle	0.55
spheres, diameter 100 nm, same total mass	1000000000000	particles	0.55

Size cutoffs decide the count

Kooi and Koelmans proposed treating particle size as a power-law distribution, so counts can be rescaled between size ranges^11,10. A 2021 analysis imaged more than 60,000 particles by infrared, down to 11 or 20 µm. Median slopes for particle length ran from 2.2 in marine surface water to 3.1 in organisms (level C)¹².

With those slopes, lowering a cutoff tenfold would raise the count about 16 to 126 times (our calculation, valid when the largest particles are far above the cutoff). When a new study reports "far more particles than before", first check whether its cutoff moved.

Whether the power law continues below the measured range is not established for environmental samples (level E). In the 2021 data the smallest particles already fell below the fitted line. The authors cite a power law down to 0.1 µm only from a laboratory fragmentation test¹². Search scope: that dataset and a web search on 3 November 2026 for field size distributions below 1 µm.

Estimates, scenarios and measurements

Headline numbers are often estimates, and estimates can be badly off. Evangelou and colleagues ran an atmospheric model with three published emission inventories against 2,782 measurements from 76 studies. The simulations overestimated measured median concentrations and deposition by two to four orders of magnitude, mainly because emissions had been overestimated (level C)⁹.

Extrapolations from one expedition carry similar caveats. Ten Hietbrink and colleagues measured nanoplastics of three polymers (PET, polystyrene, PVC) along one North Atlantic transect. They estimated about 27 million metric tons (27 Mt) in the mixed layer of the temperate to subtropical North Atlantic (estimate, level C)⁵. That is three polymers in one layer of one basin, not all ocean plastic.

Percentages: method results and global shares

A percentage only means something with its denominator. "At least 99.97% of particles 0.3 µm across" is the U.S. Department of Energy definition of a HEPA filter, as summarized by the EPA¹⁵. It is a theoretical rating at the hardest size to catch, not a field measurement (level P).

Two figures circulate for sand filtration of nanoplastics. In a pilot rapid sand filter with aged sand in Zurich, 99.5% of model nanoplastics about 200 nm across stayed in the first 0.9 m over nine hours (measured, level C)^16,17. The "more than 3 log", or about 99.9%, comes from a computer model of a full-scale treatment train, dominated by its slow sand step (modeled, level C)¹⁷. Neither says how much plastic has left the planet's air or water.

The home page shows two editorial indicators, not measured shares. "0 / 4" counts the reviewed domains (air, water, soil, human body) that meet the four conditions for a proven complete solution on Solutions by System. None did in the review checked on 3 November 2026 (level P)¹⁸.

The "≈0%" symbol on the v6 home cards had no numerator, denominator or size range. It stood for a status in words, "no proven removal at scale", and since 4 November 2026 the cards show those words instead. A local 99.5% and that status do not contradict each other, and neither says that prevention or local treatment fails.

Ranges and orders of magnitude

A range is the answer, not a weakness. Jambeck and colleagues estimated that 4.8 to 12.7 Mt of plastic waste of all sizes entered the ocean from land in 2010 (estimate, level C)¹⁹. The top is about 2.6 times the bottom (our calculation). Quote both ends, never the midpoint alone.

An order of magnitude is a factor of ten: "two orders" is a hundredfold, "six orders" a millionfold. On a log axis each equal step is ten times the last, so this site uses log scales when studies are far apart.

Worked example: the credit card a week

In June 2019 a WWF report prepared by Dalberg, based on a study WWF commissioned from the University of Newcastle, said people may ingest about 5 g of plastic a week²⁰. That is the weight of a credit card. The peer-reviewed version, Senathirajah and colleagues, gives 0.1 to 5 g per week by ingestion, for particles up to 1 mm, as a global average²¹. The headline kept only the top of the range.

Mohamed Nor and colleagues modeled a different scope: intake from eight food types plus inhaled air, for particles from 1 to 5,000 µm. Their adult median is about 583 ng per day, or about 4.1 µg per week (level C)²². Five grams is about 1.2 million times more (our calculation). The ratio compares two outputs with different scopes; it is not an uncertainty range.

Why so far apart? Senathirajah and colleagues converted counts into mass with averaged particle masses; Pletz argued that this inflates the mass by several orders of magnitude²³. Cox and colleagues, often cited alongside, counted particles only: 74,000 to 113,000 a year for Americans including inhalation, after a 2020 correction, with no mass estimate^24,25.

Run the five questions. Size range: different. Count or mass: counts converted to mass. Measured or estimated: both estimated. Samples: a few food types. Compared with what: a credit card, an arbitrary unit. The 5 g figure is disputed (level D); 4.1 µg is one model's median (level C). The site shows both with their scopes.

Mass equivalent of one week's modeled intake for adults (mg), log scale, with reference objects shown apart

Data table

label	value	unit	kind	source
Mohamed Nor et al. 2021: median adult intake, eight food types plus inhaled air, 1–5,000 µm	0.0041	mg/week	model median	S-103
Senathirajah et al. 2021: low end of range, ingestion, particles up to 1 mm	100	mg/week	low end of a disputed published range	S-104
Senathirajah et al. 2021: high end of range, ingestion, particles up to 1 mm	5000	mg/week	high end of a disputed published range	S-104
Reference object, not an intake: one 1 mm polystyrene sphere	0.55	mg	reference mass (our calculation)	site-derived
Reference object, not an intake: one credit card, as used in the 2019 headline	5000	mg	reference mass	S-105

Using this on tomorrow's headline

Ask the five questions of any new number. If size range, metric or method is missing, treat it as unreadable rather than wrong. For how instruments set size limits, see How We Measure It. For what the grades mean, see Evidence Levels. For the counter inputs, see Flows & Model; for the home indicators, Planet Overview; for intake and tissue figures, Human Body.