Industries / Food, farming and forestry / Food and beverage processing / Bottled water
Sources
Sources work is finding, protecting and holding the right to use the springs, wells and town water supplies that bottled water is drawn from.
6
Milestones achieved, from 1756 to 2024.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1756
A spring sold on purity
Dr John Wall tested Malvern's spring water in 1756 and found it exceptionally pure, with very few minerals, which gave rise to the local rhyme that it "is famed for containing just nothing at all". Malvern water was later sold by several local companies and in the end by Schweppes and Coca-Cola Enterprises.
Credit
Dr John Wall; the account is Malvern Hills District Council's
1992
Vittel pays its farmers
Nitrate from farm fertilizer had been rising in the catchment of Vittel's springs in France since the early 1980s, and French law does not allow natural mineral water to be treated to remove it. After a research programme with INRA, France's agricultural research institute, from 1989, Nestlé Waters set up Agrivair in 1992 to pay farmers to change how they farm. Farmers signed 18- to 30-year contracts to give up maize and chemical inputs and keep at most one head of cattle per hectare, in return for about 200 euros per hectare a year for five years and up to 150,000 euros per farm for new equipment. By 2004, 92% of the sub-basin was protected and every remaining farm had switched.
Lessonit took from 1992 to 2004 to win over every farm, and the farms fell from 37 to 26 between 1988 and 2006 as the rest sold land.
Credit
Nestlé Waters and Agrivair, with INRA; case study by Danièle Perrot-Maître for FAO
2016
Ontario pauses bottling permits
Ontario stopped issuing new permits to take groundwater for bottling between December 2016 and April 2021. From 1 August 2017, bottling plants have paid $500 for every million litres of groundwater they take, and the province says the charge has brought in about $770,000 a year.
Credit
Government of Ontario
2019
Pakistan's court prices water
On 12 January 2019 Dawn reported a Supreme Court of Pakistan judgment, written by Chief Justice Mian Saqib Nisar, in a case over bottled water and drinks companies selling underground water they had taken without charge. The court imposed a levy of one rupee per litre, paid into a fund for two dams, ordered meters on every extraction point within 30 days, and asked for a plan to phase out plastic bottles.
Credit
Supreme Court of Pakistan
2023
California curbs a forest pipeline
On 19 September 2023, after eight years and dozens of hearings, California's State Water Resources Control Board found that BlueTriton, owner of the Arrowhead brand, had no permission to take water from springs in the San Bernardino National Forest, and ordered it to cut back sharply. In May 2025 a Fresno County Superior Court judge overturned the order, ruling that the board had gone "beyond the limits of its delegated authority", since state law denies it power over groundwater. The board said it was weighing an appeal.
Lessonan order to protect a spring holds only if the regulator has power over that kind of water.
Credit
California State Water Resources Control Board; Judge Robert Whalen Jr
2024
US bottling plants counted
The US Geological Survey (USGS) published a national inventory of US water bottling plants, compiled in 2023, because "there is no comprehensive national inventory" that includes how much water each takes. It lists 10,192 facilities that produce bottled water. Water-use figures could be found for only 159 of them, or 1.6%.
Lessonthe inventory rests mostly on a purchased business list, because no public list of plants exists.
Credit
Carol Luukkonen, Cheryl Buchwald, Gary Martin and Allegra Johnson Mckee, USGS
Watch for
Bottling's share of lost flow
A USGS model of the Wards Brook valley near Fryeburg, Maine, for 2016 to 2021 found that pumping for town supplies and for bottling both reduced the brook's base flow, the steady flow fed by groundwater, and that about half of the pumping's effect came from bottling. Flows stayed above the recommended minimum of 2,180 cubic metres a day in that period, but fell below it in summer in modelled three-year droughts later this century. The study is one of four regional USGS studies of bottling, with others on Strawberry Creek in California, the Great Lakes states and Florida's Santa Fe River basin.
Achieved whenthe study passes peer review, or measured stream flows confirm the modelled share.
Claimed by
John Mullaney and colleagues, USGS, February 2026, in a preprint not yet peer reviewed
Open
01
Permits next to actual pumping
A public table, plant by plant, of how much water each bottling plant may take under its permit and how much it reported taking, year by year. Start where regulators publish both figures, such as Ontario, and add the US plants in the USGS inventory that have water-use records. Every row links to the document it came from.
The briefCredit · unclaimed02
Water levels near bottling wells
For US bottling plants that pump their own groundwater or spring water, outside the four regions USGS is studying, find the nearest public groundwater-level wells and stream gauges. Then compare water-level trends near each plant with trends in control wells farther away in the same aquifer, before and after the plant began pumping.
The briefCredit · unclaimed
Open 01 · the brief
Permits next to actual pumping
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table, plant by plant, of how much water each bottling plant may take under its permit and how much it reported taking, year by year. Start where regulators publish both figures, such as Ontario, and add the US plants in the USGS inventory that have water-use records. Every row links to the document it came from.
Why it is still open
USGS found water-use figures for 159 of 10,192 US bottled water plants and says no national inventory includes the volumes taken. Pakistan's Supreme Court had to order meters fitted at extraction points in 2019. A public dashboard of Ontario's permits, posted on GitHub in 2025, uses the permit file only, not the volumes reported. No table that sets permitted against reported volumes for bottling plants turned up in searches for this file.
Built with what exists
Ontario's open "Permit to Take Water" data, whose permits carry a "Bottled Water" purpose and a maximum in litres a day, and its "Water Taking" data of daily and annual volumes reported for each permit; the USGS bottling inventory and its water-use table; Python with pandas; and a small AI model to pull figures out of permit documents and company reports (see the Note "The jobs agents repeat").
First step
Download Ontario's permit spreadsheet, keep the permits whose specific purpose is "Bottled Water", and join them by permit number to the annual water-taking file for the latest year.
Done looks like
At least 200 bottling plants in at least two countries, each with permitted volume, reported volume and year, and a link for every figure. A second person checks 20 rows chosen at random against the sources and finds at most one error.
Open 02 · the brief
Water levels near bottling wells
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
For US bottling plants that pump their own groundwater or spring water, outside the four regions USGS is studying, find the nearest public groundwater-level wells and stream gauges. Then compare water-level trends near each plant with trends in control wells farther away in the same aquifer, before and after the plant began pumping.
Why it is still open
USGS writes that the effects of taking water for bottling "have not been systematically assessed, and certainly not at a national scale". Its own work models four regions in depth. At single sites outside them, the effect is argued in court: a forest group says Arrowhead's pipeline has sharply reduced the flow of Strawberry Creek in California, a claim still before a federal court in 2025.
Built with what exists
The USGS bottling inventory, which marks plants that supply their own water; the USGS Water Data API, whose "field-measurements" and "daily" collections serve groundwater levels and stream flow; and Python with pandas and geopandas for the distances and trend tests.
First step
Count how many self-supplied bottled water plants in the USGS inventory have a USGS groundwater-level well with at least 10 years of readings within 5 km.
Done looks like
A published notebook and table for at least 50 self-supplied plants outside the USGS study regions, each with the trend near the plant, the trend in control wells and the years of record behind each. Three of the plants are checked against their local hydrogeology reports.