Brewing water barely touched coffee extraction at tap concentrations

Ask a specialty shop what to do about brewing water and the answer comes back fast. Soft water goes flat. Magnesium lifts the fruit. Calcium dulls it. Buy the sachets. Behind most of that advice sits a computer calculation from 2014 rather than a brewed pot, and when a lab finally went to the bench and measured one, the result was duller than the sales pitch. At the mineral levels a kitchen tap actually delivers, almost nothing happened.

What the test actually measured

The experiment ran in Heliyon in February 2024. Brewing followed the Specialty Coffee Association protocol: 11.00 g of coffee, 200 mL of water heated to 93 °C, steeped for 5 minutes in a French press. One dark roasted Arabica blend, held constant throughout. Into the water went two salts, magnesium chloride and calcium chloride, each at a low dose and a high one.

Four organic acids came out the other side to be counted: citric, malic, lactic and quinic. The team measured them twice, once by gas chromatography mass spectrometry and once by nuclear magnetic resonance. Slow work, and hard to argue with. The salts also went in twice, before the brew in one set of runs and after it in another. That second choice turned out to carry the finding.

Ten times tap strength before anything moved

At 1000 ppm the numbers finally shifted, and they shifted hard. The magnesium salt cut citric, malic and quinic acid by 40 to 60%. The calcium salt pulled citric acid down 70% and malic acid down 25 to 40%. Lactic acid sat where it was under both, which is a clue in itself: whatever the cations were doing, they were not doing it to everything at once.

A thousand parts per million is not tap water. It is a laboratory dose, ten times the realistic concentration the same team ran alongside it, picked to force an effect big enough to see. The published results read very differently once that framing is in place, because the large percentages all belong to the extreme end.

At 100 ppm, water hardness barely moved the acids

The low dose is the one a home brewer lives with. Salts added at 100 ppm had slightly reducing or no effect on the acid content. That is the paper's own phrasing, and it is a quiet sentence for something an entire product category rests on.

Water hardness carries a lot of blame in coffee, and some of it is earned, though mostly in places that have nothing to do with taste. Scale on a heating element is real, measurable and expensive, which is why hardness sets the descaling interval on a machine. What this work could not find was that same mineral range rewriting the acids in the cup.

Before the brew or after, coffee extraction did not notice

Salts added after brewing should do nothing to extraction. The grounds have already given up whatever they were going to give up. Comparisons between pre- and post-brew additions showed similar acid content in most cases, and the authors read that as extraction of acids proceeding independent of the water composition.

That reading has teeth. If the water is not changing what leaves the grounds, the mineral effect is not an extraction effect at all, and the familiar picture of magnesium prising flavour compounds off the bean stops matching the measurements. Interactions after the brew may still shift what a drinker perceives, which is a different mechanism and does not answer to the same fix as a low extraction yield.

What that leaves on the counter

The 2014 paper the advice descends from, in the Journal of Agricultural and Food Chemistry, is density functional theory: binding energies for five coffee acids, caffeine and one flavour compound, computed rather than brewed. Good chemistry, with no pot of coffee anywhere in it. The authors of the newer work put their own position plainly, writing that a scientific basis for water quality recommendations in the coffee industry is long overdue.

So before the mineral sachets go into the basket, look up the hardness figure the local water company already publishes and hold it against 100 ppm, the dose that did almost nothing here. Then keep the edge of the study in sight. Four acids were counted. Flavour was not.