A grand unified theory of espresso

Not too long ago I posted on Home Barista about trying to find a good way to measure the density of coffee beans.  a

As always the paricipants there were way smarter than me and offered several interesting options. I dropped into the thread that this was part of my idea of a grand unified theory of espresso, and subsequently a few people mailed and pm’d me asking what on earth I was talking about and what density had to do with it.

Well, I should probably explain what I have been thinking.  b

When approaching a coffee and brewing it using an espresso machine you are often searching for an ‘ideal’ recipe for that coffee.  For many of us knowing as much about that coffee as possible often helps make intuitive judgements about things like dose and brew temp.  If I get my science wrong then please, please shoot me down.  c

One of the things that broke my head about roasting early on was the discovery that the longer/darker you roast the less potential solubles you have.  This may seem obvious to some but it always felt like the roasting process created many new flavours but one must distinguish between flavours and solubles.  This explains why instant coffee is roasting very fast and relatively light – they are interested in a percentage yield so the more solubles the better, regardless of taste.

Knowing this then made something I had experienced make apparent sense – darker roasts generally prefer higher doses, because you need more coffee to get more solubles to get a nice, thick and pleasant espresso.  But after a moments thought it didn’t make sense.  This would mean that you would get a thicker, heavier cup with a higher TDS measurement from lighter roasts but lighter roasts generally produced delicious but lighter bodied cups.  What nudged me towards density was how we approach very high grown coffees versus lower grown coffees.

Higher grown coffees (and let me make a broad sweeping generalisation here) have a much higher acidity than their lower grown brethren.  When brewed as espresso they can easily yield extremely acidic and unbalanced cups and, taking unbalanced and pronounced acidity as a sign of underextraction, I found that higher brewing temperatures helped to produce a more balanced cup.

What do lighter roasts and high grown coffees have in common:  higher densities.  Though there was more to extract (in theory) you had to work a lot harder to do it.  Therefore reducing the dose of a lighter roast/higher grown coffee gave you a higher ratio of water/energy to coffee to help extract a tasty cup.  (I often think of heat energy as some sort of currency, with which you can buy solubles.  The more heat, the more you extract/purchase.)

This is all well and good for convenient examples – high grown, light roasts versus low grown darker roasts.  Give me a clean prepped coffee from relatively low altitudes in Brasil, roast it into 2nd and I will likely be dosing quite high and not brewing too hot.  I’ll certainly be dosing it very different from how I might brew a lot of Aricha as a straight shot.

But what about a light roast of a low grown coffee, or a dark roast of something grown super high up?  This is what lead me to wanting to find a way to measure and compare the densities of coffee beans to see if there was correlation between the density of the end product and an ideal brew temperture of particularly effective dose.

So – if I have had such a good idea why don’t I do all the research and then publish it all at the end of it and try and stamp it “Hoffmann’s theory” or something equally absurd? (Apart from the fact that it is absurd).  Because I want to generate a little discussion about this.  I want people to weigh in and tell me I am being stupid/simplistic and to suggest better ways to test these ideas.  I want to understand espresso better so I can make better tasting drinks and translate coffee’s journey more transparently in the cup.  d

I really hope people will offer their opinions on this idea, shoot it down or take it and run with it.  I am going to start doing some basic testing and see what happens.  Lots of little experiments appeal.  e

  1. There really is no better place on the web for these kinds of questions!  (back)
  2. Some of this is based on personal preference, some on what seems to be fairly well agreed upon within the community of people who worry a lot about their espresso.  (back)
  3. There is another post in the works about the value of being wrong and discussing it afterwards – yes, I have a big wrongness to confess to…..  (back)
  4. I know this is a wordy post and all, and I would have put some nice photos in but my camera is dead  (back)
  5. For example – if I roast two coffees til their densities match – will they grind the same, and at the same dose will they extract the same?  Would they be ideal, therefore, to blend together to get the most out of each of them.  This probably shouldn’t be a footnote, but it is.  So there.  (back)

30 Comments A grand unified theory of espresso

  1. Kyle

    You are on the right track. Another important thing to consider is elasticity of the bean and how it translates to the grinder.

    I think the heightened perception of body and lack of clarity in dark roasted coffee has a lot to do with the way a more brittle bean will fracture in the grinding burrs. This could also be another reason for the higher instance of over-extraction in these types of coffee…

    I performed a little experiment, and I can’t say much about the consequences, but removing fines from ground espresso can have a very interesting impact on flavor…..

  2. Poul Mark

    James, in all of this, are you assuming something akin to a perfect roast? In other words is the roast being eliminated as a variable in all of this? If so, great. If not, it obviously complicates the situation, as a light roast on a SHB, and high elevation SHB is going to be more likely underdeveloped than a darker roast (more density, more time needed to develop). Underdeveloped bean, will obviously taste more astringent, more acidic, if not compensated for. Not to say that I am advocating a dark roast at all, but simply wondering what variables in this discussion are going to be considered and which one’s will not.

  3. Pingback: Topics about Espresso | » A grand unified theory of espresso by James Hoffmann

  4. Paul Stack

    Science says too many variables = MHS (Melty Head Syndrome).
    A mind map centred around the ‘perfect espresso’ with an army of em, arms, cataloguing all the significant contributors would help get the mind-muddle framed. Take altitude, roast, grind, temperature, dose, etc. – keep all but one variable static and play with the last one. Divide and Conquer.

  5. James Hoffmann

    Well I think that an underdeveloped roast would still show up with a slightly odd density for its roast colour. A higher density from the undeveloped inside would increase the overall average for the bean/grounds and you’d know to go a little hotter, tighter or lower on the dosing. In theory anyway!

  6. Nick

    I have one of these in my head that is constantly getting modified. Much of it is based on the theories that Rao lays out in his book, and most of his footnotes are from Illy. But, on a visit to the Dalla Corte factory I was told by Jens Thomsen that the reason they set out to do varying temps on groups was because singles take a HIGHER temperature than doubles. This throws a wrench in Rao’s equilibrium temperature theory, and thus in mine, because I always thought singles took lower temps. Of course, I never had the chance to use a single basket outside of quick experiments because I worked in the US.

    Basket geometry I think is something that doesn’t get much play among baristi other than “Hey, this basket works better with this coffee.” On that tour I also heard that there’s an ideal aspect ratio for the coffee column. Anything shallower and wider leads to underextracton, and I was either offered or inferred in my head that the higher doses in the US are a result of a widespread gravitation to this aspect ratio. Wider baskets need more coffee to achieve the proper depth-to-width.

    I think we’re closer to having a working unified theory than many people think; I get the impression that machine manufacturers and big(er) roasters already have much of this data.

    Don’t forget water hardness; I had an espresso here in Bologna that by all measures should have been undrinkable, but was actually pleasantly tolerable. The mineral content of the water is quite high here, which must wreak havoc on boilers but makes nice coffee.

  7. jason scheltus

    Doesn’t this tell you that lowering dose/raising temperature has more to do with the acidy taste?

    I mean instead of saying “it tastes acidy, therefore it must be quite dense, therefore I must lower dose/raise temp” can’t you just say “it tastes acidy, therefore I will raise temp/lower dose”?

    Personally, I think you are overcomplicating things. I’ve found greater changes in taste of espresso come from slightly modified airflow i.e. cleaning flues in the roaster, or ambient temperature affecting roasts. I think it’s amusing to hear of you’re thinking of introducing more variables to try and control; rather than just nailing the basic ones!

    Here’s an exercise, throw an open question out to see how many variables people can name that affect the taste of roasted coffee; as espresso or not. I’d be willing to bet there are lots and lots I had never even thought of!

  8. Pingback: Topics about Espresso | A grand unified theory of espresso

  9. ant

    Hi James,

    I think where you will really run into problems is the issue of defining the shot as the end result. We all know just how subjectively espresso is experienced around the globe and people can and will look for different flavours or textures in the same blend or even single origin. So your grand unified theory of espresso will need a way of addressing these parameters in a relational context rather than a discrete and narrow window of ‘opportunity’. Multivariate ergo non-linear fun ahead perhaps.

    How will you address flavours in relation to solubles?
    e.g. How does solubles extracted relate to getting coriander or basil as a taste out of coffee and how does this compare to solubles extracted relating to getting more red berry sweetness out of the same blend.

    There’s a lot more work involved and I’ll throw out that old quote of espresso being something fleeting, containing 8000 different taste compounds blah blah yawn and capable of seemingly infinite hairpulling antics (until of course you run out of hair but that doesn’t mean you can’t pull the hair out of people around you).

    For the record I’ve gone from updosing everything to downdosing everything… unless downdosing doesn’t cut through milk then I updose again. These days I tend to keep a little bit of space between the top of the puck and the showerscreen to allow more follow through in the aftertaste of the espresso, so in a way I’m halfway between schomer and tim wendelboe.

    ps- to clarify when I mention updosing I just mean ‘thickening the density of the puck’ but still keeping some form of clearance from the shower screen. I almost never dose till I’m touching the shower screen anymore but this is also due to a move onto machinery such as the gb5 and fb80 each with rather slow flowrates. I suspect I’d be updosing everything if I was in charge of a san marino or a la san marco.

    Downdosing as defined in my tech is usually signified by the top of my espressocraft tamper being level with the top of the rim of the portafilter basket/dosing at the middle or lowest line of stock lamarzocco baskets.

    Also, if I have a small amount of a standout coffee at home I tend to downdose and change temperature to get the required combination of pour and taste while at times sacrificing oomph. This saves coffee and allows me one extra double shot or more of daterra or blue batak or good old mountaintop.

  10. Hugo

    I might be being a little simple, but I’m wondering whether you want to know the density of whole beans or ground? Comments seem to suggest whole beans, which I understand would be relevant to roasting.
    My confusion is with extraction, as coffee is only ever extracted once ground and surely the measured density of the bean goes all to shit when put through a grinder?

    Or didn’t I understand the question?

  11. David

    James, I like where you’re going with this. I have a sneaking suspicion that Nick is right above when he says

    I get the impression that machine manufacturers and bigg(er) roasters already have much of this data.

    I was leafing through the most recent edition of “Tea & Coffee” and came across an article entitled “The Challenge of Grinding Coffee for Pods and Capsules”. It didn’t unlock any major key pieces of information, but it did give a peek into an entirely different side of the grinding and brewing equation. Most importantly, it had a photo of one of the monstrous roller grinders that I tend to pass right by at the trade shows.

    On page 81 of the magazine, there is a photo of a similar roller grinder, the “Grinder Superfine series” grinder from Probat-Werke, that was parked across the aisle from our booth at the SCAA in Atlanta. While I don’t understand the workings at all, there are two interesting features that pique my interest:

    …roast coffee producers achieve consistently reproducible degrees of density and grinding…

    …An important enhancement of the 246/1 and 546/1 models against their precursors is the automatic density control in the ground coffee densifier

    “Ground coffee densifier”?!?!?!

    Someone amongst us has to have some experience with this type of equipment, or have an acquaintance who understand how all of this works. Anyone?

  12. David

    Sorry…I am an absolute amateur when it comes to formatting. Please forgive the pitiful layout above.

  13. dsc

    Hi James,

    a friend of mine mentioned that Illy in his book states that darker roasts give higher yield (coffee weight before extraction minus coffee weight after extraction – same as brew ratio?). This would mean that more stuff dissolves or gets moved to the cup when using darker roasts. Isn’t that a bit against what you’ve written above?

    I’m guessing that darker roasts are more dry thus allow water to bond with more particles and tend to ‘give away’ more. Lighter roasts have more water inside and aren’t so happy extracting ‘properly’ (ie. less soluble). The same reason might be standing behind different grind settings. Darker roasts are dry thus crack/break easily, perhaps producing more fines which leads to coarser grind settings. Light roasts still have some water in them, so are more flexible, harder to crush, probably producing strange grind particles (size-wise) which means you have to grind finer to get proper extractions.

    Of course all of the above might be rubbish, I’m just taking a wild guess.


  14. James Hoffmann

    I think the Illy dates fits very well. Darker roasts (less dense coffees) are easier to extract – likely due to their lower density/the way they grind. They may have less solubles, but it is easier to get those solubles – hence lower brew temps with darker roasts.

    Instant coffees are roasted light, but they have to work extremely hard to get their desired yield – approx 30% – I suspect because of the roast/density.

    I have a good book, very kindly sent to me by Jim Schulman, which may have some of this data in the instant coffee chapter.

  15. James Hoffmann

    pH of what – the brewed coffee in a controlled brew (similar to a cupping) to get a baseline read? Surely the pH is influenced by the extraction itself – especially with espresso?

  16. Jaime

    It is imperative to think that the majority of variables are unable to be controlled in most roasting setups beyond maintenance and simplistic inputs anyway. The batch sizes are too large, the production too high, the roasting equipment incapable. The simple fact is it’s often easier to take a barista approach and think about how to approach the roast you are given at varying drop points than it is to change the roast profile drastically to fit a desired extraction parameter.

    Underdeveloped is in a broad generalization what we call roasting raw. You cannot compensate for that as a barista, the astringency alone can turn your stomach a bit. Ambient temp/humidity and the pipes are a solid call Jason, props on that.

  17. Josh

    Knowing this then made something I had experienced make apparent sense – darker roasts generally prefer higher doses, because you need more coffee to get more solubles to get a nice, thick and pleasant espresso. But after a moments thought it didn’t make sense. This would mean that you would get a thicker, heavier cup with a higher TDS measurement from lighter roasts but lighter roasts generally produced delicious but lighter bodied cups

    I thought the body of the espresso was due primarily to the in-solubles rather than the solubles? Perhaps later-stage developments degrade the structure of the bean in a way that really throws off the insoluble/soluble balance during extraction, which would make sense if your TDS readings are showing the way they are.

  18. Hotels Sydney

    I thought the body of the espresso was due primarily to the in-solubles rather than the solubles? Perhaps later-stage developments degrade the structure of the bean in a way that really throws off the insoluble/soluble balance during extraction.

  19. Pingback: In the (”New”) News: « coffee

  20. Pablo G Escobar

    Hi Jim,

    this is the first time that I write on this blog, and not sure if this entry is in the right place but here you go, would like to ask you and your bloggers opinion on: “should a coffe shops serve as a default a espresso in a cartoon cup as if an espresso could be drank on the move? and does it change the taste of an espresso the cartoon cup or is just my coffee snobish imagination? . I went today to my local boutique coffee shop ‘coffee plantattion’ in portobello road and I was refused to be served and kicked out! A rude imbecile who claimed to be the morning manager took offense becuase I asked him to change my espresso that he had served in a cartoon cup into a porcelain one as per usual in the afternoons when I usualy go there. Perhaps I wasn’t sufficiently polite asking the change or he have had a bad night, but great to hear your thoughts on the 2 above questions. Cheers, Pablo

  21. Andy Halterman

    ditto what Nick said about water hardness. I once switched from hard filtered tap water to bottled water with a much lower TDS on my espresso machine. With the tap water, the extraction looked thready and the shot was watery. When I switched back to the tap water, the pull was much “gloppier” and had beautiful tiger striping.

    That’s just one anecdote and not a study, but the difference in the two waters was dramatic and is probably something we should keep in mind.

  22. Benjamin Schellack


    Yesterday afternoon, while we were working on an espresso blend, I came across this post; thanks so much! It gave us the needed insight to perfect our blend!

    We cupped several (all high grown) coffees and thought some of the lighter roasted ones promised the best spro. As shots, though, we couldn’t kick a slightly acidic bite in the finish. Higher temperatures helped, but we figured we’d just have to roast darker. No! It blew my mind that “the darker you roast the less potential soluble solids you have.” Down dosing on those higher grown, higher density, lighter roasted beans really made the difference!



Leave A Comment