Coffee experiments to run at home: nine protocols

Ground rules that apply to all nine

CoffeeHQ explanation

1. Change one thing. Decide what the one difference is and hold everything else the same: coffee, dose by weight, water, temperature, time and equipment. If two things change, the result cannot be assigned to either.
2. Do not know which is which. Have someone else pour and label the cups, or mark the undersides and shuffle them. Expectation changes what people taste, and the person who made the cups always has an expectation.
3. Ask whether there is a difference before asking which is better. Use a triangle test: three cups, two the same and one different, and pick the odd one. Guessing finds it one time in three, so a single correct pick means little.
4. Repeat. Run the triangle six times. Four or more correct of six happens by luck about one time in ten; five or more, about one time in fifty-six. That is arithmetic, and it is the difference between a hunch and a finding.
5. Write it down before you look. Record your picks and descriptions before the labels are revealed. A note written afterwards has already been edited by the answer.

A result from one coffee on one day is true of that coffee on that day. It is evidence about your own set-up and no more.

1. Does the alkalinity of your water change the cup?

CoffeeHQ explanation

The one controlled tasting study of water found alkalinity changing a cup far more than hardness. This asks whether the difference between two waters you can buy is one you can taste.

1. Hold constant. One coffee, one grinder setting, one brewer, the same dose and water weight, and both waters heated in the same kettle to the same temperature.
2. Do. Choose two bottled waters whose labels state bicarbonate, one low and one several times higher. Brew the same recipe with each, side by side, and run the triangle test.
3. Record. The bicarbonate, calcium and magnesium from each label; your picks; and, once revealed, one or two words for acidity and bitterness in each.
4. What limits it. Bottled waters differ in everything at once, so a difference cannot be put down to bicarbonate alone. Tap water from the same supply varies through the year.
5. How to read it. If you cannot pick the odd cup reliably, water is not what stands between you and a better brew at these two levels. If you can, the lower-bicarbonate cup tasting sharper would agree with the published study; anything else is worth repeating.

2. Does the filter material change the cup?

CoffeeHQ explanation

Paper is said to give a cleaner cup and metal a heavier one. What has been measured is that paper holds back oil. This asks whether you can taste the difference in one brewer.

1. Hold constant. A brewer that accepts both a paper and a metal or cloth filter, one coffee, the same dose, water, temperature and pouring pattern.
2. Do. Rinse the paper. Brew one batch with each filter, time both, and run the triangle test on the cooled cups.
3. Record. Total brew time for each, since the filters will not drain at the same rate; whether sediment is visible; your picks.
4. What limits it. A different filter changes the drain time, so contact time changes with it. Sediment and an oily surface are visible, which undoes the blinding unless you taste from opaque cups.
5. How to read it. A reliable difference shows the filter matters in your brewer. It does not show whether oil, fine particles or the changed brew time is responsible.

3. What does sieving out the fines do?

CoffeeHQ explanation

A grinder makes a spread of sizes. This uses a kitchen sieve to change the spread without changing the grinder, and asks what happens to drain time and taste in a pour-over.

1. Hold constant. Coffee, grinder setting, brewer, filter, water and temperature, and the weight of grounds that goes into the brewer.
2. Do. Grind more than you need. Shake half through a fine tea strainer or sieve to remove the smallest particles, then weigh out the same dose from the sieved and the unsieved grounds and brew both.
3. Record. The weight of fines removed as a share of what was sieved; the drain time of each brew; your picks in the triangle test.
4. What limits it. A kitchen sieve has an unknown mesh, and sieving takes minutes during which ground coffee loses gas and aroma. Keep the unsieved half waiting for the same time.
5. How to read it. A faster drain without fines is what a porous bed would lead you to expect. Whether the cup is better is a separate question, and in the one published test of added fines in espresso a single taster, who was not tasting blind, found no penalty for them.

4. How does your espresso flow, second by second?

CoffeeHQ explanation

A shot time is one number for a process that changes throughout. This records the whole curve with a scale and a phone.

1. Hold constant. Coffee, dose, grind, distribution, tamp and machine settings across at least three shots.
2. Do. Put the cup on a scale under the portafilter and film the scale's display from the moment the pump starts. Afterwards, read off the weight every five seconds.
3. Record. Time to the first drop; the weight at each five-second mark; and from those, the grams gained in each interval.
4. What limits it. A kitchen scale lags and rounds; the first seconds are the least reliable. Three shots are few.
5. How to read it. Lay the three curves over one another. How closely they agree is a measure of how repeatable your preparation is, before any question of taste. A published study found flow rising through a shot as the coffee dissolves; a curve that jumps suddenly is one to compare with how that shot tasted.

5. Does tamping force change your shot?

CoffeeHQ explanation

The two short statements found on this in research papers report no appreciable change in shot time or extraction across a range of tamp pressures, and no detectable effect on how fast two compounds were extracted. This tests it on your equipment.

1. Hold constant. Coffee, dose to a tenth of a gram, grind, distribution, yield by weight and machine settings. Keep every tamp level.
2. Do. Make six shots, alternating a light tamp that just settles the bed with the hardest you can comfortably repeat. Pressing the tamper on bathroom scales beforehand gives a rough figure for each.
3. Record. Shot time to the target yield for each of the six, and the two approximate forces.
4. What limits it. Six shots is a small sample, a hand is not a calibrated press, and the grinder's own shot-to-shot variation is in the data too. A lighter tamp also leaves the bed standing higher in the basket, so the space above the coffee changes along with the force.
5. How to read it. Compare the gap between the light and hard averages with the spread inside each group. If the spread within a group is as large as the gap between them, force is not what is moving your shots.

6. Does needle distribution make shots more alike?

CoffeeHQ explanation

Stirring the grounds with needles before tamping is described as necessary for reproducibility by people who do it, and no controlled comparison was found. Reproducibility is something a scale and a timer can measure.

1. Hold constant. Coffee, dose, grind, tamp, yield by weight and machine settings.
2. Do. Make ten shots, alternating: five stirred with the needle tool, five simply levelled by tapping the basket.
3. Record. Shot time to the target yield for each, in order, and anything visible under the basket if the portafilter is bottomless.
4. What limits it. Ten shots on one day with one coffee. A grinder that clumps badly and one that does not may give different answers.
5. How to read it. Look at the range of shot times within each group of five, not at the averages. A tool that evens out the bed should narrow the range. Taste the two most different shots of each group.

7. How does one bag change as it rests?

CoffeeHQ explanation

No study was found in which a panel tasted one coffee day by day after roasting, and a resting window on a bag is the roaster's judgement. This follows one bag.

1. Hold constant. One bag with a roast date, one recipe, one grinder setting to begin with, the same water, and the bag resealed the same way each time.
2. Do. Brew the same recipe on fixed days after the roast date, for example days 2, 5, 9, 14 and 21, and write notes each time before reading the earlier ones.
3. Record. For filter: how much the bed rises when first wetted, and the drain time. For espresso: shot time at the unchanged grind setting. For both: a few words on taste.
4. What limits it. This cannot be tasted blind or side by side, because the days are different days, and memory of taste is poor. The physical records are firmer than the tasting notes. Opening the bag each time lets air in, so this follows an opened bag and not a sealed one; and the room's humidity and the grinder's temperature differ from day to day as well.
5. How to read it. That gas leaves roasted coffee over weeks is measured chemistry, and a shrinking bloom is usually read as that. A shot time that drifts at an unchanged grind setting is usually put down to the same cause, which nothing read has tested. Your notes show when you liked that coffee, which is not a rule for the next one: darker roasts hold more gas and lose it sooner.

8. Do frozen beans grind differently?

CoffeeHQ explanation

One study found colder beans giving a narrower spread of particle sizes, and tasted nothing. This asks whether the change shows up in a brew.

1. Hold constant. One coffee, split into two sealed portions from the same bag on the same day; grinder setting, dose, recipe and water.
2. Do. Freeze one portion overnight in a sealed container and keep the other in a cupboard. Grind the frozen portion straight from the freezer and the other at room temperature, and brew both the same way at once.
3. Record. Drain time or shot time for each, and your picks in the triangle test.
4. What limits it. Condensation forms on cold beans in a warm kitchen, so grind the frozen portion immediately. Check the grinder maker's instructions first; some advise against it.
5. How to read it. A different drain or shot time at the same setting shows the grind changed, as the study would predict. A change in time alters the brew, so a difference in taste may be the time and not the temperature; adjusting the grind to match the times and tasting again separates the two.

9. What does milk temperature change?

CoffeeHQ explanation

Steamed milk is widely said to taste sweetest in a middle range and flat or cooked beyond it. This uses a thermometer and no coffee.

1. Hold constant. One carton of milk, the same starting temperature and volume in the jug, and the same amount of air let in at the start.
2. Do. Steam three jugs to three temperatures read on a thermometer, for example 50, 60 and 70 °C, and pour a small glass of each. Taste them once they have all cooled to the same warm temperature.
3. Record. The temperature each reached, the weight of the jug before and after to see how much water the steam added, how the foam looks after two minutes, and a few words on sweetness and any cooked taste.
4. What limits it. Sweetness is perceived differently at different temperatures, which is why the glasses should be tasted at the same one. A hotter jug has also had more steam and so more water added.
5. How to read it. A difference that survives cooling to the same temperature is a change in the milk. One that disappears was a change in how warmth is perceived. The weights give your machine's own figure for dilution.

What these protocols cannot do

CoffeeHQ explanation

They report no results, because CoffeeHQ has run none of them, and they predict none. Where a published study bears on a question the protocol says what that study found; a reader's result may differ, and on their own equipment the reader's result is the better evidence.

They are not science in the sense of the studies cited elsewhere on this site. One taster, a handful of cups and a kitchen scale cannot establish a general fact. What they can do is replace a rule taken on trust with an observation, and show how much of the difference between two cups survives when the taster does not know which is which.

None of them involves modifying equipment or doing anything a maker's instructions advise against. Where a protocol touches on something a manual may cover, such as grinding frozen beans, the manual comes first.

What to do next

• Tasting Lab — A triangle test with the odds of guessing worked out, and a worksheet for notes that stay in your browser.
• Espresso puck physics — The evidence behind the flow, tamping and distribution protocols.
• Water chemistry and brewing — The tasting study the water protocol follows, and why alkalinity is the property to compare.
• Why coffee goes stale — What is measured about gas and resting, behind the one-bag protocol.
• Grinding and particle size — Why a grinder makes a spread of sizes, and what colder beans did in the one study.