Methodology
Where the numbers come from.
The catalog is the product, so this page is the receipt.
Every mineral value on this site traces to a published analysis, and every water stores which kind: the producer’s own label or analysis certificate, an accredited laboratory report, or an official register (state mineral-water registries, municipal water reports). That source tier travels with the data — you’ll find a “Verified · official analysis” link on each water page that goes to the actual document, not to us.
Only verified waters can win a match. Waters we can’t yet source to that standard stay in the catalog marked unverified, and the matching engine holds them back from the top spot. When a value simply isn’t published — silica is the usual culprit — the page says “not published”. We never estimate, average, or fill in a plausible number. A missing value is information too.
The gate
Before anything ships, a validation gate re-derives every water’s classification from its raw minerals and refuses to build if a stored value disagrees. It also runs hydrochemical sanity checks — ion-balance closure, stated TDS against estimated dry residue — and prints warnings we review rather than hide. Some published labels fail those checks; where they do, the warning stays on record until a better analysis surfaces.
What we’ve corrected
Public water data is messier than it looks, and we fix it in the open. A few examples from our own audit log: Oslo tap water’s magnesium circulated at six times the real value — the municipal lab report says 0.4 mg/L, so that’s what we print. A Swiss spring’s values were widely copied from a summary that no published analysis actually supports; the official register disagreed, and the register won. Several “metasilicic acid” figures needed converting before they could honestly be called silica — a ×0.769 unit trap that quietly inflates numbers when skipped. And more than one brand’s numbers turned out to be a decades-old label superseded by the producer’s current analysis.
Our magnesium “bitter edge” line is a taste threshold, not a label threshold: we set it at 100 mg/L because that’s where reported perception begins for sensitive palates (Health Canada’s magnesium review; the average person needs several times more) — not at the EU’s 50 mg/L “contains magnesium” marketing line. Where a claim and a flavor differ, we side with the flavor and say so.
Citing WaterVibe
Cite a water page directly — for example “WaterVibe: Gerolsteiner, mineral analysis (mywatervibe.com/w/gerolsteiner)” — and it will keep pointing at the current sourced values; slugs don’t change. Machine-readable versions of the catalog live at /llms.txt for an overview and in the per-page structured data. If you spot a value we got wrong, tell us — being corrected in public is the whole point of publishing sources.