Cupping Protocols And Extraction

I often find myself contemplating whether we're focusing on the right priorities. Tonight, I made a decision to conduct some straightforward and simplified experiments to examine the impact of our coffee cupping processes.

In our usual cupping routine, we begin by weighing the whole coffee beans, grinding them, and then pouring just-off-the-boil water. After allowing it to steep for about 4 minutes, we break the crust that forms on the surface. Following this, we skim off any residual foam and grounds while the coffee cools to a suitable tasting temperature. The tasting process typically continues until the coffee approaches room temperature.

This may end up being a fairly long post so you can skip down to the ultimate conclusion of the whole thing.

My friend and I stayed after work tonight and ran three little experiments. The first was mostly a failure due to bad method. The idea was to track how quickly a cupping bowl was extracting. For the sake of replication – or more likely correction – here is the method we used:

Experiment 1 Method

• We used 12g of coffee per bowl to maintain consistency. To eliminate potential grinder retention variations, we first ground 70g of coffee, and then measured it into the individual bowls.
• Each bowl received an equal weight of water (with minor fluctuations due to recorded human error), aiming for a target of 200g.
• The water was sourced from a custom Uber boiler tap, exiting at a temperature of 97°C.
• Filling the bowls was staggered to enable measurements at specific time intervals: 30s, 60s, 120s, 240s, 300s, and 420s.
• To measure, we sampled 5ml of liquid from just 1cm below the crust. This sample was filtered into a separate vessel and allowed to cool before being measured using an Extract Mojo device.
• Variations in brewing water weight were taken into account when calculating the extraction.
• The same coffee was used for all bowls ((I’m not going to spam you with the coffee name)

Experiment 1 Results

This was something of a failure as the timings had not been worked out well, and some timings markers were missed. I would consider this experiment pretty much garbage but we did get the following results:

Time (s) |	Extraction %
30s	14.5
50s	14.7
135s	17.55
180s	17.6
240s	17.6

Experiment 1 Conclusions

As you can see the results don’t match the planned method. This is a bit embarrassing and I plan to run this again and organise myself a little better. Shame on me…

At best I am going to offer hypothesises that can be tested further by other people.  It would appear that most of the extraction happens very quickly in the first 30s in an infusion type brew, and then the rate of extraction seems to decrease and then plateau.  Coffee does not continue to extract to the point of overextraction if you leave it a long time without breaking.  I don’t really find this to be valuable data but I’d love other people to have a go at this and see what results they get.

These results are more interesting in the context of the next couple of experiments.

The next experiment was about the break in the context of time.  How important is it that you break a bowl at a consistent time?

Experiment 2 Method

- 12g of coffee per bowl as before.

- The bowls were filled with an identical weight of water as before.

- Water was exiting a short, custom Uber boiler font at 97C.

- Filling was staggered to allow the break to take place at a desired range of times: 30s, 60s, 120s, 240s, 360s & 480s.

- Two measurements would be taken:  One when all bowls had reached 10 minutes of brewing, and then again at 20 minutes of brewing.  The reason for this was to see if extraction would continue after the break.

- Variations in brewing water weight were factored in when calculating extraction.

 Experiment 2 Results

In the end the extraction data was so similar at 20 minutes to the data at 10 minutes that is isn’t worth posting. Here is what happened:

Time Broken |	Extraction %
30s	17.45
60s	18.24*
120	18.03
240s	18.65
360s	18.61
480s	19.08

* This bowl had an accidentally higher dose of water in it – 212g instead of 200g. This is relevant when looking at experiment 3.

Experiment 2 Conclusions

For me this was really interesting.  Even when you got stuck into the break after only 30s you still had a reasonably extracted cup, and waiting a long time didn’t mess things up either.  Most people are breaking around the 240s mark (though obviously with their chosen grind/extraction preferences).

Most shocking to me was that the break really does seem to stop the coffee extracting further.  The TDS of the cups measured 10 minutes after this had barely changed – by 0.01% in most cases.

Secondly the bump in extraction due to breaking is notable.  This was the same coffee and grind setting as the previous experiment.

It would seem that being accurate to the second isn’t hugely important in brewing coffee, and that there is a fairly large window in which you can break and not have a large effect on the coffee.

Problem:  Each bowl was skimmed clean after breaking so we could taste after sampling.  This meant that some ground coffee was removed from each bowl, so not all bowls had the same amount of total grounds in them for the duration of their brewing.  How much this impacted the resulting strength/extraction is unknown.

Experiment 3 Method

There is a lot of focus on the break for many people, but there is rarely a focus on the weight of water used.  We wanted to run a simple experiment to see how the ratio of coffee to water would effect extraction.

- 12g of coffee was used in each bowl.  It was weighed ground as before.

- For each bowl we varied the weight of water added, from 170g to 220g in steps of 10g.

- Each bowl had an identical break time, an identical skimming clean time and we also measured the temperature of each bowl at 7 minutes using an IR thermometer.

- Samples were taken at 11 minutes for each bowl, filtered and tested.

- Different coffee was used in this experiment to previous ones (explaining the change in extraction range).

Experiment 3 Results

Brew Water Weight |	Extraction
170g	14.26
180g	16.14
192g	16.56
200g	17.16
210g	17.28
224g	17.77

Represented graphically:

Here are the temperatures we recorded at 7 minutes of brewing, for each weight of water:

Brew Water Weight |	Temperature
170g	58°C
180g	61°C
192g	63°C
200g	64°C
210g	65°C
224g	66°C

Experiment 3 Conclusions

The Extent Of The Relationship Between Brew Water And Extraction In An Infusion Was A Little Shocking To Me. If This Had Been A Percolation Then I Would Have Been Less Surprised.

The Temperature Was An Interesting Side Effect, I Guess Down To Thermal Mass. All Bowls Were Identical Temperatures At The Start Of The Experiment, And Were Taken From Shelves At Room Temperature And Had Not Been Used In Previous Experiments.

It Would Seem Likely That The Increased Extraction Is A Combination Of Increased Solvent Present, As Well As More Energy In The Form Of Brewing Temperature. It Implies That Mass Of Water Is Perhaps Of More Concern In A Cupping Environment Compared To Brewing Times And Break Times.

Important: All Bowls Are Not Created Equal. Nor Do All Coffees Bloom Equally. I Would Be Confidant That Most People (Including Ourselves) Have A 10g Swing In Either Direction Of Their Average Brewing Weight When Cupping. This Means A Fairly Large Swing In Their Extractions, Which May Be Influencing Their Roasting Or Purchasing Decisions For The Worse.

Overall Discussion

Some could, and maybe will, argue that they don’t worry about brewing weight, never have, and they’ve found their cuppings to be useful and effective QC or buying tools. I would say that it feels like we go to great pains in certain aspects of cupping – such as the weight of the coffee itself being targeted to 0.1g – but we ignore other hugely important variables. We have a cupping setup that would allow identical water temps and in theory identical weights of brew water. I don’t mind admitting that we haven’t zeroed the scale on every single bowl we’ve cupped. There is, however, a lot of received wisdom and ritual around cupping and I’d like to strip some of it back.

My final disclaimer is that the above isn’t very good, nor methodical, science. I hope it encourages a few people to experiment and test out the effects of their own protocols. It has certainly made me rethink what we prioritise when we are cupping and fixing our method that we teach and train.

I’ve missed doing geeky stuff on here. I’d love to see people post their own experiments on their blogs, improve my methods (shouldn’t be hard!) and see what results they get.