The topology of the coffee extraction and strength landscape remains rather poorly charted, and pocketed with pitfalls. In searching for and as yet not finding documents that sufficiently described the methodologies originally employed by the Midwest Research Institute (and other bodies) in collecting and tabulating preference data we have revealed that there are more questions about how exactly that data, and our coffee control brew charts were created. That isn’t to say I have grave doubts over their validity, certainly my own palate tends to mostly agree with their optimal ranges. Questions remain though. How many samples did they collect? What coffee did they use? To what degree was it roasted? How was it ground? What was the grind profile?
Grinders in particular remain to me the area whose potential for improvement could most dramatically increase the quality of our cup, but also have economic benefits derived directly from improved grind profiles. I spoke recently in Berkeley on this subject and I tagged the talk with the phrase:
Bad grinding costs money, wastes coffee and produces a mediocre cup.
To explain this we must return to the topology of the extraction and strength landscape. When we measure strength with (most likely these days) a refractometer and infer extraction we come to an average extraction, or a macro extraction. Looking more closely, however, in the cup we have a population of extractions whose spread is dictated by many factors including even wetting, bed geometry, agitation, but for the purposes of this discussion in a major way it is dictated by the grind profile.
When I say grind profile, I am referring to the distribution of particle sizes in a sample of ground coffee. The physics of bean breaking, or at least how we currently break them, and the brittle three dimensional hexagonal lattice that is roasted coffee, equates to the inevitable formation of sub-particles (fines). Different burr sets, with diverse tooth patterns, and of varying sizes produce grind profiles with more or less uniform grind size distributions.
For espresso percolation, the conventional wisdom is that a bimodal or plurimodal grind profile (that is one with two or more peaks) is preferable in that it permits the formation of a more densely packed bed, with limited porosity that can provide resistance against water under pressure. The extent to which this is true and not just a happy coincidence with grinder technology is a debate for another day, but let’s assume it is true.
For filter coffee (and by filter I include not just drip, but all brewed coffee methods from press to siphon to Aeropress and beyond) we are not as constrained to create a coffee sample with this kind of mechanical consideration. In filter coffee we can more easily adjust factors like contact time, bed geometry, temperature etc to compensate for the ability or inability of a coffee sample to oppose the flow of water. In other words we can hypothetically employ a completely uniform grind profile.
This hypothetical grind profile is currently just that – hypothetical. All grinders, even the best ones produce an uneven grind, or at least a not-completely-even grind. The best ones produce a more-even grind. To look at it another way, they produce a cup whose average extraction has a smaller standard deviation. To know the grind profile of the coffee used to collect the original preference data might inform how we interpret that data.
How do we choose a desired extraction? We choose the point at which we extract the most desirable flavours with the least impact of negative flavours. I tend to find myself around 19% in general (with a Mahlkonig Tanzania at home), but it can depend a little on the coffee in question. The negative flavours of over-extraction, accumulating bitterness and astringency define the point at which I say stop. However, as our extraction is an average, only a certain proportion of the grind population is contributing to these flavours.
The smaller particle sizes, sub-particles, fines will reach maximal extraction and have begun to diminish the cup quality before the larger particle sizes have given up all their desirable flavour compounds.
As the sub-particles, or secondary peak is often a much smaller part of the overall grind, a minority, this means that the minority is having a disproportionate impact on the point at which we cease extraction. The main peak, or that which is defined by the grind setting (i.e. the separation of the opposing burrs) in a good grinder makes up the majority of the grind with a narrow spread of particle sizes – it is limited in its extraction by the point at which the smaller particles begin to dominate.
In delivering this topic in Berkeley I brought with me two coffee sieves, one at 700um and one at 800um, which allowed me to laboriously isolate a grind sample with no fines and of a spread of just 100um. To illustrate my point, I brewed this over 7 minutes at a 50g/L ratio in a V60 reaching 24% extraction and about 1.38% strength. I served this to those in attendance. The cup belied its extraction. While it had an average extraction of 24% it (to all extents and purposes) had nothing beyond that, so the cup was not marred by bitterness or astringency – it tasted great.
This also means that I reached a final beverage strength that had I targeted 19% extraction would have required an extra 10g/L coffee. In other words with a completely uniform grind I could use less coffee and extract more – saving money and waste – and produce a better cup.
When I first played with sieving coffee a couple of years ago, and even in last year’s Aeropress championships this point was lost on me. I was still targeting 19% extraction, because that was what I had come to expect with a normal grind profile. While it tasted great, clean and sweet, everyone who tasted those cups with me concluded it was missing something. This also means that given a completely uniform grind the region of interest on the coffee brewing control chart would almost certainly shift to the right (UG = Uniform Grind below).
In a smaller way it might mean that depending on how the original coffee used by the MRI (and the other bodies who carried out this type of research) was ground, our grinding for the SCAE Gold Cup research may be conceivably more or less even, and could result in a preference shift to the left or right as a result.
It also for me confirms the trend for up-dosing and under-extraction and lays the blame (at least partially) at the feet of poor grinding. The more uneven your grind the sooner you will want to stop extracting, the more you compensate for this by increasing dose. If you are brewing filter using an espresso grinder, which (for the most part) will have a bi/pluri-modal distribution this may well be what you are doing. It is not only producing a poor cup, but it is wasting coffee (and money).
Grinding technology despite some refinements, some improvements and optimizations has not significantly advanced in half a century or more. Does this stagnation or dark age mean that what we have is good enough? It may mean that the economic and cup-quality benefits of a more-uniform or completely-uniform grind will remain for the foreseeable future – hypothetical.