A coffee can taste pointedly sour and oddly hollow at the same time. The tempting answer is to grind finer, keep the water in contact longer, and call the result “more extracted.” Sometimes that helps. Sometimes it adds bitterness without bringing back the fruit or sweetness you expected. Before moving the grinder again, it helps to ask what, exactly, was uneven: the water’s passage through the grounds, the mixture of substances in the cup, or your perception of that mixture.
These are related questions, but they are not interchangeable. Extraction yield is the total mass of coffee solids dissolved from the dry grounds. It is an average. It does not say whether one patch of a coffee bed gave up much more than another, nor does it tell you which acids and bitter compounds ended up in the drink. A refractometer can estimate the concentration of dissolved coffee in the finished cup; your tongue encounters a chemical and sensory balance that one number cannot describe.
First find out where the water went
In a pour-over or espresso bed, water follows paths through a packed mass of particles. Those paths need not share the work evenly. Kevin Moroney and colleagues modelled flow through cylindrical and cone-shaped coffee beds, checking part of their model against experiments. In the cone, their simulations found local differences in flow and extraction: a single final extraction percentage hid a range of extraction levels inside the bed. This was a model with simplifying assumptions, not a photograph of every domestic brew. It gives a useful warning about what an average can conceal.
A later two-pathway mathematical model of espresso explored how a small difference in bed permeability could grow as extraction changed the pathways themselves. It was offered as an explanation for the counterintuitive observation that grinding ever finer can eventually lower overall extraction. The model is a plausible mechanism, not proof that a particular puck channelled. Still, if you see an erratic drawdown, a visibly uneven wet bed, or large swings between otherwise similar brews, the physical route of water deserves attention before its mineral content does.
Water chemistry asks a different question. Every drop passes into the coffee carrying dissolved substances of its own. Two waters can have the same total dissolved solids and quite different proportions of calcium, magnesium and bicarbonate. Calcium and magnesium are the principal contributors to hardness. Bicarbonate contributes strongly to alkalinity, the water’s capacity to resist a fall in pH when acids enter it. A reading that rolls all dissolved material into one total cannot tell you which of these you have. The original work on dissolved cations by Christopher Hendon and colleagues makes clear why the identities of ions matter more than a single sum.
The attractive theory, and the experiment that tested part of it
Hendon’s team used quantum chemical calculations to compare how sodium, calcium and magnesium ions bind to a small set of coffee-related molecules. Magnesium bound most strongly in those calculations. It was a reason to investigate whether magnesium-rich brewing water would pull more flavor compounds into the cup. It was not a sensory trial or a direct measurement of what a real brewer extracted from roasted grounds. The researchers themselves called for experiments on the larger-scale effect.
One such check came from Tove Bratthäll and colleagues in 2024. They added calcium or magnesium chloride either before brewing or after brewing, then measured four organic acids—citric, malic, lactic and quinic—using two analytical methods. At concentrations typical of drinking water, the acid measurements changed little. When salts were added at ten times that concentration, the numbers varied more, and the analytical methods did not always agree. Crucially, adding salts after brewing often produced results much like adding them beforehand. That comparison weakens the claim that the ordinary calcium and magnesium levels in this experiment substantially changed how those four soluble acids left the grounds. It does not settle what happens to every other coffee molecule, with every roast, or in every brewer.
The important lesson is the size of the claim. A calculated attraction between one ion and one molecule is chemically interesting; a cup of coffee contains many molecules competing in hot water, and the measurement itself can respond to the matrix. “Magnesium extracts more” is too broad a kitchen rule for the evidence. The narrower finding—little effect on four measured acids under the tested conditions—is more useful because you know where it stops.
What bicarbonate changes in the cup
For taste, bicarbonate may be the more consequential number. In a 2026 controlled sensory study, Christina Birke Rune and colleagues brewed a Colombian coffee in French presses with prepared waters that varied calcium, magnesium and bicarbonate. Trained assessors did not find a consistent sensory change from hardness alone across the tested waters. Higher bicarbonate, however, made the coffee seem less acidic and less fruity, with more roasted and bitter character. The brew’s pH moved in the same direction, supporting an acid-buffering explanation.
The result needs its scale. In the study’s discrimination tests, bicarbonate at 120 and 240 milligrams per litre produced detectable differences from the ultrapure-water reference; 15, 30 and 60 milligrams per litre did not do so consistently on their own. A separate descriptive test included 300 milligrams per litre, an intentionally high experimental level. The sensory work centred on one Colombian coffee and one brewing method. The authors also measured pH with other coffees, but they did not repeat the full tasting panel for every origin and roast. It would be a mistake to turn this into a universal target number or a promise that any calcium adjustment is pointless.
There is another subtlety. Less perceived acidity need not mean that less acid was extracted from the grounds. Bicarbonate can alter the acid balance after the compounds enter the liquid. The 2026 study measured sensory attributes and brew pH; it did not map all the compounds extracted under each water condition. In a separate 2023 study of acids in brewed coffee, Rune and colleagues measured several acids across five coffees and varying light roast levels, then tested how readily selected acids could be detected or recognised. The relationship between the concentration of a named acid and what a person calls “acidity” was far from simple. That is a good reason to describe the cup precisely—muted fruit, sharp sourness, drying bitterness—rather than assigning every sensation to a single molecule.
A small test worth making
If a familiar coffee suddenly tastes flat, compare two safe drinking waters with known, different bicarbonate or alkalinity values. A water supplier’s analysis or a bottle’s mineral label is more informative for this question than a total-dissolved-solids reading alone. Keep the coffee, dose, grind, water-to-coffee ratio, brewer, temperature and contact time as steady as you can. Brew both close together; taste them at similar temperatures, ideally without knowing which is which. Repeat on another day before treating a preference as a finding. This is a kitchen comparison, not a laboratory isolation of every mineral, but it can tell you whether changing water is worth pursuing.
If the flavor changes consistently while the brew runs similarly, investigate the water profile. If flow and extraction vary from cup to cup on the same water, inspect grind, wetting, distribution and pour pattern. Both can be happening at once. A strongly buffered water may flatten bright notes while an uneven bed produces its own mix of under- and over-extracted regions. Repeated grinder changes cannot tell those causes apart unless the water is held still for the test.
There is no need to chase a single “perfect” mineral recipe. Beans, roasting, brewing method and taste preference all change the question. Machine requirements and scale risk also matter, especially for espresso equipment, and a taste test cannot settle those. The useful skill is to distinguish a path through the grounds from a reaction in the cup. When the next brew tastes uneven, make one change that tests one explanation. The result will teach you more than an extra turn of the grinder made in frustration.
A thought worth keeping?
