The espresso grind setting where finer stopped extracting more

A shot runs thin, so the barista grinds finer. It is the oldest reflex behind a bar, and a 2020 paper in the journal Matter took it apart with a mathematical model and a year of shots pulled in a working cafe. The researchers tracked how much of the dry coffee actually ended up dissolved in the cup as they walked the espresso grind setting from coarse to fine across the whole usable range of one grinder. The line went up. Then it came back down.

How the shots were actually pulled

The recipe was deliberately boring: 20 g of dry coffee in, 40 g of beverage out, every single time. The coffee was an espresso profile specialty roast from Supreme Roasters in Brisbane. What would not stay still was the espresso machine. At 9 bar the fine end of the dial simply clogged, and a clogged shot cannot be measured, so the water pressure came down to 6 bar for the whole experiment. Lower pressure and a lighter tamp together opened the range back up, and the dial ran from 1.1 to 2.3 depending on the coffee.

That pressure drop was made to reach the fine settings without choking the basket, not because a softer pump tasted better, and there is a longer argument about why 9 bar became the default at all.

Where the espresso grind setting turns over

Two things move as the burrs open up and only one of them behaves. Shot time falls in a straight line: coarser grounds, less resistance, a faster shot. Nobody argues with that one. The second relationship is the one worth taping to the grinder. Extraction yield, the share of the dry dose that ends up dissolved in the drink, did not keep climbing as the coffee got finer. It rose, crested, and fell away again, with the decline setting in below a grind setting of 1.7.

Measured yields across the range sat inside a narrow band, roughly 21% to 25% of the dry mass, which is why the turn is nearly invisible without a refractometer and several hundred shots. The model predicted the opposite: smaller particles mean more surface and shorter paths out, so the maths said finer should always win. At the fine end, reality stopped agreeing.

What partially clogged flow tastes like

The explanation the authors settle on is uneven flow. Pack a bed tightly enough and it stops acting like a uniform filter. Water finds the slackest path, races down it, and leaves the rest of the puck barely rinsed. The figure coming off the refractometer is an average over all of it, so the number falls even while parts of the bed are being pounded.

That has a taste, and it is one most people have met. A single espresso reading as bitter and sour at the same time, two flavours from opposite ends of the chemistry, sharpest as the cup cools. The published work also shows how little a lone number settles: at a fixed brew ratio, two shots both measured 22% extraction yield, one at a grind setting of 1.3 and one at 2.0. A 20 second shot and a fast one cannot be carrying the same molecules, which is a standing caveat on any extraction figure read off a chart.

Tamping pressure barely moved the needle

One ritual at the bar had to be ruled out first. A range of tamping pressure was tested and produced no appreciable variation in how long a shot ran or how much it extracted, so the variable was retired: every puck in the study was pressed at 98 N by a Puqpress, an automated tamper accurate to within 3 N.

Flavour differences were noticed and never quantified, so this is not permission to put the tamper in a drawer. What it says is narrower and more useful. The force in a barista's wrist is not the thing holding their shots apart, and the coffee grinder is doing far more of the work.

Less coffee, coarser grind, the same result

Here is the move the paper actually argues for. Cut the coffee dose from 20 g to 15 g, grind markedly coarser, and the same yield returns with far less spread between shots. Those shots can finish in under 15 seconds. The drink comes out less concentrated and tastes different, so it is a trade. The reason almost nobody works this way is written into the rulebook: the Specialty Coffee Association puts espresso at a 20 to 30 second extraction, and chasing that window pushes a barista straight into the settings where the bed starts to choke.

One cafe in Eugene, Oregon has run the protocol since 2017. Its till between September 2018 and September 2019 counted 27,850 espresso-containing drinks, each of which used to take 20 g of coffee valued at $0.53. Pulling 25% of that mass out saved $0.13 a drink and $3,620 over the year, and the shots settled at roughly 14 seconds.

So the next time an espresso tastes thin, try the wrong direction. Open the grinder a notch instead of closing it, take two or three grams off the coffee dose, and watch the clock rather than the dial. If the time drops and the cup does not thin out with it, the water was never reaching most of what the basket held.

Which part of the shot is still yours to choose?

Grind, dose and pressure are all dials, and the research keeps finding that they move the cup less than the ritual around them suggests. What goes into the hopper is not a dial at all. Santo Cafe roasts a single origin Mexican arabica for exactly this kind of brewing, and the pack says plainly what is inside it: Mexican coffee, 100% arabica, freshly roasted. The grind can be argued about afterwards.