by Professor Steven Abbott; Mark Abbott; Sebastian Abbott

Acknowledgements: We thank Dr Anja Rahn for her Count Rumford articles^[1] that directly inspired us to do something closer to the Count’s optimum. And we thank Anja and Barista Hustle for lively debates along our journey from laughably failed initial experiments to coffee we really enjoy brewing.

Here we describe our UX – Uniform Extraction – method for brewing great coffee. But first, some context behind what we set out to do.

To a scientist the filter cone brew method (one example is the popular V60) is crazy. It’s an “out of control” system where everything depends on everything else. In an ideal method you can “separate the variables”. For brewing this would mean that temperature, grind size, extraction time, clarity/fines-level of the final drink, can all be controlled independently. These cone brews have many rituals (pour the water 3 times clockwise…) that exist to overcome the fact that at just about every part of the process, everything is changing in time and location. It really is out of control. Sure, if you try really hard you can dial in a reliable drink, but science says that separating the variables would mean less work for better results.

Looking in another direction, to non-experts such as ourselves, the French press has always seemed to be a method scorned by those who are expert, despite the wonderful video from James Hoffmann showing that in some ways it is superior to the cone brew, provided you can accept the absence of clarity in cup. But when you compare our goals for a controlled UX brew, below, with Hoffmann’s method for the perfect French press with its mysterious crust breaking at 4 minutes, you will see that it, too, lacks separation of the variables. There are some ingenious YouTube hacks for the French press, such as adding filter paper to the plunger, but, again, they don’t separate the variables…

And although it might seem heresy to question the wisdom of the cupping method, it seems to us to be a bizarre way to get to the essence of a coffee’s capabilities when it is such an out-of-control method. Don’t take our word for it, Liang et al from USC Davies^[2] point out the inherent problems of “concentration gradient inversion” during the process. Yes, we understand the convenience of cupping but, no, we don’t think it’s an acceptable method and suggest near the end how to do things better.

Whether espresso is more or less out of control doesn’t concern us here as our target is a “normal” cup of coffee, though we have made a few delicious espresso-style drinks to show that it can be done with little effort.

So we set ourselves the mission of creating a brew method that:

• Is simple to do, with just one, affordable, new-to-the-coffee world accessory;
• Is repeatable;
• Lets you choose your degree of roast;
• Lets you choose whatever grind size you prefer;
• Lets you choose your brewing temperature and extraction time;
• Naturally keeps you close to that brewing temperature during your chosen time;
• Exposes every particle of coffee to the same extraction environment;

• This makes it a Uniform Extraction method, hence our name UX

• Has a fuss-free and scientific way to separate (most of) the grinds from the brew;
• Makes it easy to choose your final mouth feel;
• In honour of Count Rumford:

• Minimizes aroma escape;
• Minimizes convection during a key part of the process;

• In summary, lets you dial in the coffee that you want, via a single general-purpose technique.

To a large extent, this is a list of “separating the variables”, so it’s a scientific method. Because it is basically an immersion technique, we need to point out that neither the French press nor the Aeropress, not even the latest fancy variants around filter/immersion, naturally meet the key requirement to expose “every particle of coffee to the same extraction environment”, nor is separation of the grinds “fuss-free and scientific”. The same criticism applies to cupping.

As we shall see, for those who like near-zero mouthfeel we hit the problem that confronts everyone – the fact that fines block filters. The ways around it are rather interesting as we’ll find later.

The basic method is easy to describe. For simplicity, we call the brewing vessel the “brewer”:

1. Grind the coffee to the size you like
2. Heat the water to the temperature you want
3. In a double-walled insulated brewer (with a good lid), weigh in your water, get it stirring with a magnetic stirrer, tip in your coffee then continue with the magnetic stirrer for your chosen brewing time
4. Let settle for your chosen settling time
5. Pour (gently) via any simple filter that gives you the mouth feel you want – and catches any sludge
6. Enjoy

Here’s the setup:

We admit that there are is a compromise: the settling period allows some out-of-control extraction, but we’ve ensured that settling is as fast and controlled as physics allows.

First buy a magnetic stirrer

From your favourite online source you will have no difficulty buying a magnetic stirrer for, say, $US40. It will be a small plate with a speed control. The most basic will work, but if you can pay a little more for a stronger stirring magnet and a digital speed control, life will be easier.

At the same time buy a few stirrer bar magnets at a few $ each – the bigger/beefier the better as long as they can fit nicely into your brewer. If you are going to use different vessels, you may need to play with a few different sizes. These stirrer bars are low cost, safe to use in glass vessels and are familiar to anyone who’s worked in just about any lab.

And don’t forget to buy a magnet retriever stick to pull the magnet from your brewer or (yes it happens) from out of your drain if it accidentally goes in.

Now get a double-walled glass brewer

The double-wall not only keeps you at your chosen temperature longer, it also reduces convection in the later stage. Even more perfect would be a vacuum flask and, of course, people like Gordon Howell have used these in the past with championship success in their own specific techniques. The advantage of the double-walled glass brewer is that you can at all times see what is going on. As we shall see, temperature stays reasonably steady and convection is greatly reduced, so we’re happy with our choice. Feel free to go full vacuum if you wish!

Finding the perfect brewer isn’t as easy as we would have liked. In the UK we used the 350 and 450ml ecoBrew double walled cups with an excellent lid. They work well but the 450ml is too small to make two full drinks, and there is no pouring spout which made experiments a bit tricky.

Double-walled vessels are not expensive, and those we found were all from the necessary heat-safe glass. It’s just that they aren’t too common. We’ll assume that the ingenuity of the coffee community will ensure a good supply of optimised brewers.

We strongly recommend that you get used to your combination of stirrer, magnet and brewer using just water up to your chosen the level. Stirrers work better when they are centred above the magnet in the plate – but how accurate do you have to be? It’s easy to find out. Then how reliably can you get the stirrer up to some initial stir rate – it’s hopeless to try to go fast right from the start. Finally, how fast can you spin it, with how big a vortex, and how fast would you like it to go? On one of our stirrers, at 1100 rpm we can get a vortex right down to the stirrer bar. This is not what you want for making coffee, but it’s good to know in advance what your stirrer/bar/vessel/volume combination can achieve. Just be aware that with the coffee, stirring is somewhat harder … and if you forget to turn the stirrer on, and leave the coffee to settle, a weak stirrer/bar combination might not be able to get things moving, so you will need to dig down with your spoon to get things going.

And another tip: avoid handling the brewer below its centre line. Finger marks get in the way of observing a key part of the process.

We’re ready to brew

Grind your chosen weight of coffee to your chosen size. Although most discussion on particle size distribution say “there’s no direct link between a sharp particle size distribution and the taste of the coffee”, here we can say that the link is direct – for reasons we shall see shortly. So a better grinder is preferable. But don’t worry, the original experiments were done with a simple hand grinder and the delicious results (after some initial failures!) encouraged us to refine our method and switch to a higher quality electric grinder.

Now get ready with your scales, double-walled brewer, magnet and a simple spoon. Heat your chosen water to your favourite temperature.

Pre-heat the brewer, then weigh in your hot water, get it mixing on the stirrer then add the coffee to the vortex. We initially added water to the coffee (feel free to experiment both ways) but we preferred this way.

With light and medium roasts, pouring into the vortex produces only modest amount of CO2 bubbles which quickly disappear, so you get good mixing, and no spoon is required. With dark roasts some spoon work helps. There are three reasons for the dark roast being a bit harder to get into the vortex:

1. The dark roast has more CO2 so there is more gas to bloom
2. The dark roast particles have a lower density so can more easily float to the surface on the bubbles. The density differences between roasts have a further profound effect on the process, discussed below.
3. The dark roast has more oil so is harder to wet.

Remember to put the lid on as soon as possible to capture aroma, to preserve water temperature and to set up for a stable settling environment.

While the stirrer is stirring at a moderate speed, you are extracting efficiently as all particles are exposed to the same flow conditions. The laws of physics are with you.

When do you start your timer? We recommend that you start your timer when you bring the water and coffee together. Although the first, say, 30s are a bit chaotic while things come together, the grounds are being mixed vigorously so you can argue that brewing started when the coffee enters the water.

The settling phase

After your chosen brew time, switch off the stirrer. Leave the lid on to keep the internal environment stable – this will help with the settling. Although the circulation continues for a while, there will be very little relative motion between coffee particles and the water, so extraction will quickly slow down. It’s not a complete halt so we don’t have perfect control, but it’s pretty good. If anyone were to make a specialist magnetic stirrer for this method, it would control the spin-up rate, maximum rate and stirring time, and then reverse the stirring for a few seconds to stop the rotation sooner. We found that it was simple to slow the stirrer in the last 15 seconds.

Now we need the particles to settle to allow easy pouring.

Here we use Stokes Law, which you can explore in this app: https://www.stevenabbott.co.uk/practical-mechanical/Settling.php.

The graphs show the position of particles (P) settling from the top and starting from half way (P½) then show the equivalent values for your fines.

The app assumes that your water has a density of 1 g/cc and a viscosity of 1 cP. The Stokes formula then tells us that the speed at which the coffee particles settle depends on the density of the particles, and on the square of the particle diameter. The time to settle then depends on the overall height. The three other parameters include inputs to the Boycott effect described below.

A detour into density

What value should you put into Density? For a dark roast, 1.05 seems to work OK. For a medium it might be 1.1 and a light roast might be 1.2. This implies that the light roast settles 2x faster than the medium which is 2x faster than the dark. This (and the difference in amount of bloom) agrees with the cupping observations in the Liang paper – fast settling with a little bloom for light, slow, with a stubborn bloom for dark. We, and they, don’t have precise density values in water because the situation is complicated. Although plenty of papers show that dark roast beans have lower particle density and higher porosity than light roast, confirming what we know about expansion of beans during roasting, we have not found a way to go from known porosity to density of the particle in water, as this depends on many factors.

These differences in behaviour are important. Presumably these effects depend on the age of the coffee. We were using “fresh” roasts and had no old beans with which to compare them.

Now you have been alerted to the differences, you can actively explore the different behaviours in your different roasts. We were happy that the lighter roasts that tend to be more popular work especially well with this method thanks to their rapid settling.

Back to the method

With your given density, control of settling time depends on your choice (and quality) of grind size. If you halve the grind size, you quadruple the time needed to settle. In our experience, most of the settling is “fast”, over in ~ 1 minute. The real problem is that any particles that are, say, half your set size will take 4x longer to settle.

Now you see why a good grinder is important. A sharp main peak and a small tail of fines will get you settled quickly.

At first it was hard to monitor the settling in the murky liquid. But if you get a bright flashlight and if the glass is clean (fingerprints scatter the light!) it’s easy to see the settling process.

Because the smaller particles fall so slowly, you don’t want to do anything that encourages them to move upwards or sideways. And that means we use the double-walled brewer and lid to minimize heat loss from the surface and via evaporation. This automatically minimizes convection currents. As Dr Rahn showed in her article and video, an uninsulated, open vessel experiences strong convection currents. Using, as she did, amber in salt water, we failed to see any significant convection in our system.

As a bonus, the reduction of cooling by convection, conduction and evaporation ensures that your brewing temperature remains relatively constant. In our standard 250ml setup the water poured at 93 °C (into an unheated brewer), just after stirring in the coffee was 87 °C, after 1 and 2 min stirring was 86 and 85 °C. When settling it was at 84 and 83 °C in the 3rd and 4th minutes after start. Temperature is impressively under control. This confirms that you can brew at your chosen temperature after some simple setup tests.

If you really want to speed up the settling, then tilt your brewer, of width b at an angle θ of, say, 30°. The app then calculates the extent of the amazing Boycott effect, allowing you to easily halve the time needed to settle. See the app for an explanation and a link to a YouTube video that shows what’s going on.

However good your grinder, there will always be some particles at ¼ the diameter that will take 16x longer to settle … and so on to even smaller particles. At best they provide wonderful mouth feel, at worst, they are easily filtered out in the final step, unless you want cone-filter clarity in which case we have to switch technique.

Your chosen mouth feel

As long as most of the sludge/silt has settled (which you can easily verify with your flashlight), you can enjoy your chosen mouthfeel very easily. Pour gently out of the brewer so you’re not sending out too much sludge, and a normal tea strainer will catch any stray big bits and warn you of the first sludge as you get near the end of pouring. The drink will have maximum mouth feel. Not your taste? Put a piece of “tea filter” paper into the tea strainer, and with careful pouring you get a significant reduction in mouth feel, without hitting problems of a blocked filter.

Couldn’t we have come up with a more elegant solution? We had all sorts of ideas, but the tea strainer idea won because of its simplicity.

The higher density of the light roast means that you see little sludge near your pouring spout. Sludge appears sooner for the lower density dark roasts .

If the clarity is still not good enough, if cone-filter clarity is your goal, we need some workarounds, since any good filter is immediately blocked by the fines, which are poured out before the larger sediment particles can help to pre-filter them. See below for our solutions.

What about … ?

What about cleaning and composting? Throw anything from the tea strainer back into the brewer and add plenty of water. Use the magnet extractor stick to stir up the sludge and pull out the magnet, then rinse them under water. Pour the contents of the brewer out through a slightly larger strainer/filter to separate the grinds for composting. Wash with a little soapy water and giving a wipe-down gets it clean again – inside, of course, but outside, too, for better viewing of the settling.

What about TDS and Extraction? One of the many things we love about the technique is that with the same setup we can cover a wide range of brew styles, while ensuring at least the possibility of good flavour. What do we mean by “at least the possibility”? We don’t know in advance how to dial in, say, the optimum espresso-like strength without harsh over-extraction or bitterness, but our 40g of 20:60 espresso-like drink wasn’t at all bad. The laws of physics say that if we chose to focus on improving it, we could. And we created a weak French press drink, because we could. The majority of our drinks were our agreed standard of 15 g of coffee for 250g of water along the lines of a cone filter or Aeropress or what we imagined a really good French press might be like. Our typical TDS was ~1.4%.

For fun we tried doing things with cezve fine grind and 10:100 ratio. Doing it this way isn’t as dramatic and stylish as a real cezve, but has a reassuring reproducibility. Even without the boiling that’s part of a cezve, 1 minute stirring is more than enough to achieve good extraction. Settling for, say, 4 min allows a relatively sludge-free pour, though a tea strainer/filter combination helped to get most of the drink with minimum sludge in the cup. We obtained a TDS of ~2.2%. Yes, there’s delicious mouthfeel and some small amount of sludge, but in repeated tests it proved to give reliably good results, at least to our taste.

We discuss over-extraction later, using the new Extraction app for insights.

Thanks to Dr Rahn and coffee trainer Helena Oliviero we created an app to do the standard brewchart things around dose, TDS, %Extraction etc.: https://www.stevenabbott.co.uk/practical-mechanical/Coffee-Calculator.php. Hopefully others will find it useful.

What about expert tasters? Naturally we’ve asked the Barista Hustle team to carry out their own investigations and they will report their findings.

For home brewing, the method seems to be easy, reliable and adaptable. But can this method enter the commercial world and fit into a busy barista’s workflow while meeting other requirements of cost and sustainability? Again, that’s for the experts from Barista Hustle to comment on.

Back to the Cone Filter

As mentioned above, with a not-very-fine original grind, even after settling for some time, if we poured (sludge-free) into a cone paper filter it blocked in seconds, making it impossible to get a clear cup of coffee. As is well recognized, one reason the pore-clogging fines are kept away from the cone filter paper is that the larger particles act as a pre-filter bed.

In principle, a brewer with some sort of built-in filter system at the top could be used – stir from the bottom as usual, invert to allow rapid settling onto the filter by the bigger particles then, somehow, open the filter tap to get your coffee. With the equipment to hand we got a partial success, but it went against the whole idea of essential simplicity.

Eventually we hit upon one solution that worked surprisingly well, at least for dark roasts.

Have your cone paper filter (or metal filter if you prefer) ready, pre-heated/wetted with some hot water as per your preference. At the instant your brewing time is ended, and while the coffee is still well dispersed, pour quickly into the filter.

This is a super-fast, in-control clear drink which we all greatly enjoyed. Instead of the filter trying to do many things badly, the brewer gives the desired control over temperature and extraction, with your choice of dose and grind size, while the filter does what it does best which is to use the filter bed as the main filter, with the paper providing necessary backup to those fines that make it through the bed.

We quickly learned 6 tricks:

1. Due to the volume of liquid being transferred from the brewer to the filter in a short space of time, it is best to use at least a 2-cup filter;
2. As the mixing is only 1-2 min and we have no need for controlled T after mixing, a single-walled brewer can be used. We did our experiments in old French presses – without their lids/filters;
3. To avoid the magnet splashing into the filter, either pull it out with the magnetic retriever, giving it a swirl to avoid any settling, or use a strong magnet underneath the brewer to trap it;
4. A lot of aroma and plenty of heat can be lost in the chosen 1-2 min. To reduce the losses, place a light, flat sheet on the top of the brewer. Similarly, you lose aroma and heat from the filter, so after pouring, place the sheet on top of the filter;
5. Light roasts settle so quickly in the brewer that it is hard to get the bulk of the grounds into the filter before the fines block it. No doubt there are ways around the problem which we’ve not yet found;
6. The sixth trick, described below, reverts to the original settling technique thanks to the use of a finesweeper.

Extraction time: the app

We’ve created drinks from grinds varying from super-fine to coarse. Because we can choose temperature, time, mouth feel, and have rational ways to deal with fines, we can make drinks that typically require different techniques and equipment. Given this versatility, in a grind size range from, say, 100μm to 1000µm, what might be the optimal stirring time?

The thoughtful analyses by Moroney et al from U Limerick^[3], were nicely confirmed by Wang and Lim from U Guelph^[4], allowing us to find a good-enough equation to describe coffee extraction’s dependence on particle size and brewing temperature. A Weibull two-parameter equation has a rate constant k and a shape parameter β, assumed to be constant. We can take k from their 93 °C “fine” extraction with particle size D of 643 µm and adjust it over a range of T and D values.  You can very quickly see the impact of changing T, D or both. Here we replicate their results, based on a 1:12 ratio giving 2% fully extracted TDS. Note the very fast initial extraction to over 50% of their final value.

https://www.stevenabbott.co.uk/practical-mechanical/Coffee-Extraction.php 

Although we haven’t measured particle size, we typically worked in the “medium-fine” range of our grinders, with 1.5 min stirring. Within the limits of our refractometer, we had obtained “full” extraction. The fast paper filter method, with its ~ 2min of filtration, gave a good result in less than 4min. Those who prefer the medium mouth feel, settling for 3 min before pouring through a tea strainer/filter combination, gives a good result after a total of 5 min.

As we don’t have trained palettes, particle sizing, range of coffees, precision refractometers, we leave refinements of the technique to those who have.

What about over-extraction? There is very little good literature on the topic. We found just one paper^[5], in Korean, that did a careful analysis of volatiles that appear in over-extraction. Others point out that the astringency is from non-volatiles such as chlorogenic lactones and/or partially soluble macromolecules. Whatever the cause, the fast initial extraction noted above gives us a puzzle. If, as the app shows and is observed in most studies, over 50% is extracted in the first seconds, surely this part of the coffee must over-extract while the rest of the coffee is extracted normally. What is this 50%? Papers commonly say “it’s the fines plus the shattered surface of the larger particles.” Our own take on the topic is that because there’s no instant cut-off of extraction during the settling phase (the rate is much reduced but not to zero), it’s better to use a slightly larger grind size requiring, say, 2 min extraction so that slowing circulation in the ~30s initial settling time is not causing serious over-extraction. Although we could deliberately over-extract by stirring for, say, 5 minutes, we didn’t find problems when stirring in the 1-2 minute range.

And if, as is often stated, the astringent materials are semi-solids, then the larger particles descending during settling may be enough to gently re-capture them. The finesweeper may also help.

Using a finesweeper

Suppose you could insert a finesweeper (a succinct, fun word) into the brewer, something that could catch and hold on to the fraction of particles (this might be 5% by weight for a medium grind) that cause so much trouble when you want to filter. This would eliminate those particles, so that a relatively simple filtration of the remainder would give reasonable clarity, and a metal or paper filter could be used, without blocking, for those who wanted greater clarity.

We found that a filter paper or a suitable fabric held in place around the inside of the brewer had no problem capturing these fines. The stirring now accomplishes two tasks – extraction and driving particles into contact with the finesweeper surface, where, by the laws of physics (see https://www.stevenabbott.co.uk/practical-adhesion/partical-removal.php), the smaller particles get stuck. With the usual time to settle, we could then filter with the tea strainer or with our choice of cone filter.

Note that the finesweeper isn’t trying to “filter” – there’s no liquid passing through to block it. It is simply using physics to capture the small particles at the fibre surface where fluid flow (thanks to the “no-slip boundary condition”) is minimal.

At the time of writing, two of us use a strip of folder filter paper, while one uses a strip of interfacing; in both cases they are trapped in place under the lid and act during the brewing. They both reliably sweeps up lots of fines. The filter papers can be disposed; the interfacing is easy to wash out afterward. Those who use cloth filters know how hard they can be to clean, but the interfacing is a coarse fabric which is not a problem to rinse clean. We’re sure that the coffee community has the talent to find a more elegant matching of double-walled jug and finesweeper insertion, making sure that anything used is food-safe and free of extractables that could mar the coffee flavour.

The simplicity of UX

Although we continue to look for ways to refine the process, the fact is that each of us has ended up with a fuss-free, simple, reliable method to generate uniformly extracted coffees we like to drink. Each step has its own clear logic and function and we’re not fighting the laws of physics. Although we’re impressed with those top YouTubers who can master complex techniques, we like the fact that we can keep it simple yet can, if the mood takes us, make tasty brews from cezve strong to French press weak with just a few tweaks to our daily routine.

Scientific cupping

Suppose there was suddenly a good supply of small, USB-chargeable magnetic stirrers with automatic timers and a reverse swirl option. They could probably be designed, built and shipped in a few weeks given modern tech capabilities. Now do all cupping tests with standard cups, with a small magnetic stirrer bar and good lids. Even extraction is guaranteed. There are none of the “phase inversions” noted in the Liang paper, so the hidden problems of roast level variation disappear. Settling time could be made objective, especially if double-walled cups were used to reduce convection currents.

Such a development would be especially interesting because cupping and this day-to-day brewing method are essentially the same. Expert tasters would taste what drinkers will drink.

The nature of coffee science

A final comment on the process of getting here. In retrospect, the core method is simple and obvious. But going from Dr Rahn’s inspiring Count Rumford articles to the final method was anything but simple. It involved failed experiments, dumb ideas, arguments, discussions, more failed experiments … and the occasional flash of inspiration before we got here. This is how real science works. There are two key ingredients for success. First, don’t avoid failure (that’s impossible) but learn from the failures. Second, have a small team of people with different skills, different mindsets who are willing to absorb different ideas, try new things, and keep going till the outcome is acceptable to everyone in the team.

[1] Dr Anja Rahn, Brewing the best coffee according to science, Barista Hustle, (2025) https://www.baristahustle.com/research-papers/brewing-the-best-coffee-according-to-science/
[2] Jiexin Liang et al, Concentration gradient inversion via sedimentation of coffee grounds in full immersion brewing, Journal of Food Engineering 357 (2023) 11619 https://doi.org/10.1016/j.jfoodeng.2023.111619
[3] Kevin M Moroney et al, Coffee extraction kinetics in a well mixed system, J. Math in Ind, (2017) 7:3, DOI 10.1186/s13362-016-0024-6
[4] Xiuju Wang and Loong-Tak Lim, Modeling study of coffee extraction at different temperature and grind size conditions to better understand the cold and hot brewing process, J Food Process Eng. 2021;44:e13748, https://doi.org/10.1111/jfpe.13748
[5] Jin-Sung Lee et al, Analysis of Off-flavor Compounds from Over-extracted Coffee, KOREAN J. FOOD SCI. TECHNOL. Vol. 43, No. 3, pp. 348~360 (2011)