There’s a lot we know about grinding. Grinding finer helps us extract more, grinding more coarsely speeds up the flow rate of an espresso. We know that a hot grinder under heavy use changes the grind size and the quality of the bed of ground coffee. The analogy of the title, however, refers to everything we don’t know. The results we can taste, watch and log, but which we can’t fully explain or understand, much like dark matter. And in some cases we question whether or not the ideas we’ve posited even exist. In fact, this analogy carries nicely across a lot of concepts in coffee. We can infer something’s existence and its significance on the quality of our coffee. We can taste the difference, in some cases we can attain more evidence to support and improve our understanding of the difference we taste. But at other times we are left to try and draw correlations and causation from anecdotal evidence and speculation. All concepts start their life speculatively and become stronger and more tangible if they can be tested and proven. The nature of a lot of our thought about coffee means that we change our thinking regularly as we test ideas and find them to be unsubstantiated.
The pursuit to perfect coffee and to understand what goes into it illuminates different elements for us to investigate, with certain elements attracting more attention and consideration at different moments in time. Grinding is definitely a ‘hot topic’. Moving beyond a time where all the equipment chat was mostly about a gleaming hot water dispenser, all be it a very advanced and beguiling one. It is increasingly recognised that the grinder is at the heart of coffee preparation and its arguably the biggest tool at the Baristas disposal.
For a long while I have been under the impression from industry wisdom that a more uniform grind size was the goal, we believed that the narrower the bell curve the better the extraction as it is more even and predictable. But we also knew that because roasted coffee is quite brittle, that it will likely break to produce tiny fragments, fines, and slightly bigger pieces, so-called boulders. I remember watching a fascinating presentation by David Walsh where he employed a unique approach to solving this problem. He soaked beans in water before grinding them to achieve a less brittle bean and then ground the coffee to a more even distribution. I don’t know if he actually achieved a completely uniform distribution, I’m sure it was more even, but it would be interesting to know if it was all, for instance, 100 microns.
His conclusion as far as I interpreted it, was that this changed everything, so much so that the coffee didn’t taste like itself. Extraction percentages that we see as correct or desirable become possibly less relevant as the base from which the number was derived from has dramatically changed.
The starting point for the consideration of what grinding means is rooted in our understanding of solubility, of how the water extracts compounds (and the resulting flavour derived from these) out of the coffee and into the resulting solution- a cup of coffee. Voila
The logic being that the compounds and resulting flavour groups are extracted in 4 speculative groups, each group decreasing in solubility. The first, most soluble fraction, being fruit acids(light weight acids) and organic salts, closely followed by Maillard derived light aromatics. Third would be sugar browning by products and lastly heavier organic matter, usually these taste woody, toasty and heavy etc.
Following on from this we can then surmise that an under extracted coffee suffers sensorially as we have an imbalance of more first group flavours versus less last groups and vice versa for over extraction.
This understanding is in my opinion at the core of brewing good coffee. With the barista trying to understand and utilise how all the variables affect this extraction.
It then follows that a dose made up of big pieces (boulders)of coffee will likely under extract and a dose made up of tiny piece of coffee(fines) will over extract. Therefore a dose of varying sizes will – all else remaining constant- result in varying types of flavour extraction making up the total extraction. The question of what constitutes a boulder or a fine is not clearly answerable. At least I don’t think it has been defined. Is a fine anything under 90 microns(a micron being1/1000 of a millimetre) and a boulder 120 micron and up, is there an in-between of normal grind’s?
The fines will extract more easily and the water will be more likely to move through all groups on a fine. Whereas the boulder will extract less speedily during the same time/conditions, relative to its weight. The extraction must then be stopped before the fine moves on to giving up unpleasant heavy compounds.
A narrower range of distribution should then allow you to hit a higher refractometer reading and retain a well extracted flavour.
All else constant, we would assume that this should allow us to brew identical flavoured coffee made with exclusively fines or exclusively larger pieces/boulders and achieve a similar extraction through modulating extraction time whilst maintaining the heating energy. Theoretically you should have extracted both coffees to the same point in term of dissolved solids and the evenness of the particle size is the defining characteristic as opposed to the size itself.
So far, so good.
But, what if difference size grind particles act very differently to each other during the extraction?
There are a lot of other possible questions to consider, that go beyond this. The above explored concepts assume a lot about the physical properties/nature of different size grinds within a single dose. It assumes the particles have a uniform chemical composition. But what if certain organic structures in the bean are more likely to produce a fine or boulder? Then a grinder that produces less boulders could make the organic material more likely to become a boulder more soluble by breaking it up more. This may or may not be true…. And requires a few experiments, which we hope to do. This blog is a think piece and some of this will prove completely incorrect.
It is more likely, I think, that the difference between grind sizes dictates the heat transfer properties of particles. For instance, smaller particles reach uniform heating before larger ones. This increases the accessibility of dissolving the group 3 and 4 compounds because their solubility is dramatically affected by temperature. How different is this heat difference for larger coffee pieces? Well it is defined by the relationship between volume and surface area. Small bits have a very high surface area to volume ratio, big particles have a much smaller ratio, so predicting how these particles interact with temperature is challenging. Temperature of extraction has, as we know, a huge impact on flavour, look at a cold brew extracted to the same TDS reading at the same ratio as a hot brewed coffee.
If this is happening, then the most desirable grind isn’t just a narrow distribution of grind size, but its possibly a narrow distribution of a specific size of grind, one that provides an optimum heat transfer and resulting type of extraction. This would mean a desired brew time as well. For example, if a small size was best, then brew times would need to stay short and a different recipe that achieves the same extraction with a bigger size wouldn’t actually achieve the same extraction as the physical interaction between the water and coffee has changed, possibly dramatically. This would also feed into the discussion of possible differences between similar extraction percentages achieved via espresso and filter methods.
A lot of this follows from some brewing experiments we did with sieving for different sizes and then adjusting other brewing parameters to achieve the same extraction, this was hard to mimic. With larger/boulder based extractions we needed to provide application of heat throughout the brewing process as the longer brew times needed to get good extraction from larger pieces would have very different temperatures variation throughout the brew. With fines driven cups, it was difficult to not over extract, the coffee needed to be separated from the water almost immediately. I would say the tests were not anywhere as rigorous or controlled as they could have been. But those drinks fed heavily into this blogs thought process.
Similar extractions at various sizes didn’t taste that similar at all, maybe the fines driven cup needed better filtering, or the sieves resulted in a variation of sizes- an illusion that we had achieved different uniform sizes, when really we’d split the particles up into different bandwidths of unevenness. When sieving, it would be interesting to analyse the results and see how uniform the fines we have sieved off actually are. Would the larger size sieves give more variance amongst boulders or do fines stick to boulders, or are there extra fine fines in the fines sieved dose.
But, all the same it does beg the questions posited above. If different sizes do different things then we may be looking for something different from our grinders. There’s no doubt the EK43 achieves high extractions that taste good. Is this just because the grind distribution is narrower, or is it doing something else as well?
Chris (Mr Hendon whom I worked with on water) has a theory that if the distribution of size in the dose is non uniform the heating will affect the extracted elemental composition of each particle based on size and heat capacitance. That is, there may well be a desirable ratio between fines and boulders that determines optimum extraction.
This may be true. For example the fines cup I made tasted weird, it was both heavy and highly acidic. It didn’t taste typically over extracted. And theoretically it should have achieved a higher extraction that tasted good as the grind was more uniform. The extractions I liked weren’t that high.
The uniformity aspect is very tricky to assess, especially when sieving, For instance, if one were to grind some coffee on the EK for instance and then sieve off the fines, they have eliminated a whole load of one size, thus making the resulting coffee dose of more uniform size, right?
The EK produces a hell of a lot of fines, and when you swap the measurement on the grind assessment graphs form total volume (that is, the total volume that that particle size equates to within the date) to surface area, the fines spike becomes the dominant and strongest part of the graph, in fact the fines represents the biggest amount of uniform and narrowly distributed particles within the overall dose. when you consider that they will also be responsible for the majority of a un sieved extraction things get interesting. The most narrowly distributed part of the dose has been taken away, the resulting dose is now in many ways more unevenly distributed, especially as far as extraction is concerned.
The term “evenly distributed” can itself be full of semantic problems. A narrow bell curve – a lot more of one size could be argued as very un evenly distributed in comparison to a completely flat one that had an even number of each and every particle size, from fines through to boulders. Bit of a word game really. Its fair to say though that in coffee, we mean most of the dose being one size. Even then though, it makes looking at graphs a bit mind bending. You could have a narrower distribution from fines to boulders, but the distribution could be less or more even than a wider one.
This whole topic is a bit mind bending to be honest.
If there is a difference in organic make up from fine to boulder on average then I wasn’t actually even making the same coffee with a uniform fines size at all. I was in fact making a part of that coffee, Id sieved off certain parts of the coffee. This would then make sieving a very different experiment to Mr Walsh’s, as his achieved a uniform grind within the same dose. If I achieved a uniform set of fines from a starting dose compared to sieving, I may get very different results. Maybe.
For arguments sake, let’s say this isn’t happening and that the fines cup is representative of a cup made with uniform size, if this is the case, then something else is happening.
Maybe with no larger bits around, the heat means we quickly extract elements we don’t want from the fines as the heat starts higher and stays higher for the duration of the extraction. Maybe we then want a larger size that takes some of this heat, allowing the top end of the extraction of the fines to take place at a lower temperature. In essence this may mean there is a perfect ratio between two sizes of grind.
The EK definitely produces a lot of two different kinds of sizes. Its overall distribution does appear to be narrower but it produces a lot of fines. (Noted here as approximately 0-80 microns)When the axis on the commonly distributed grind profile assessments is changed from total volume, to surface area contribution, or amount of individual pieces, the graphs then looks very different. It also produces a lot around size 150-400microns.
Is it this relationship that contributes so much to its success, along with a lack of boulders*. The last question I have regarding boulders and fines is to do with extracting from used material, such as pulling a shot through a used puck. Will a boulders surface be more heavily mined for solids by the water? If its un able to get deep in to the boulder will we take different compounds from the surface to reach 20%. As opposed to the fines, where the heat transfer means that we extract different compounds by moving through more “fresh” un mined material to reach the same extraction percentage.
A good friend of mine, won’t like this article, it asks far more questions than it even attempts to answer. I kind of like that. Questions for the sake of it however aren’t always useful, but nagging uncertainties need exploring, and the pursuit to answer them is worthwhile, even if they are dismissed. We are always learning about coffee. I’m excited by the curiosity our industry does and can have, and I look forward to looking back on this at some future point, hopefully benefiting from a new understanding about the mysteries of grinding.
* boulders present an interesting concept as regards espresso flow rate. Logically a finer grind will produce a slower rate and vice versa, the classic analogy being the flow of water through sand, as opposed to the flow of water through pebbles. I guess the question though, is would the odd pebble chucked in with the sand actually slow it down further as the sand almost holds the pebble. The water goes found it and us presented with a smaller space of sand particles to then pass through. Are the pebbles/boulders effectively creating a number of bottle necks within a dose.