One of the commonly discussed variables in espresso preparation is distribution techniques. There are many techniques and methods for distributing the grinds in the portafilter basket but which one truly gives the better espresso? We decided to conduct a little experiment to find out for ourselves once and for all. Which is the King of Distribution Techniques?
Since taste is subjective(a little) and we don't want to get into a debate about subjective matters, we will be dealing solely with numbers in this experiment.
A quick low-down for the readers out there, coffee is roughly 28-30% soluble. Out of this 28-30%, only about 20% is desirable while the rest doesn't taste very good. Calibrating an espresso is pretty much all about finding the optimum point in your espresso extraction where you can get the highest possible extraction of desirables while keeping the undesirables to a minimum. This is done through manipulating the many variables in espresso making, one of which is distributing the coffee grinds in the portafilter basket. In order to find out how much of the coffee we have extracted, we first use a coffee refractometer to find out the total dissolved solids (TDS%) from our coffee and cross reference it with a formula to calculate the final extraction yield (EY%). Here's a nice little video about how to use a coffee refractometer:
Now that we've got the basics outta the way, let's dive straight into the experiment!
Disclaimer : We are by no means statistical / math / science experts. This is just a simple experiment carried out to give us an idea of the performance of each method only. People are more than welcome to give their feedback on how we can improve and carry out a more academically approved experiment / report.
Ethiopia, Sidamo Guji
Process : Sundried Naturals
Variety: Ethiopian Heirloom
Roast Age: 10 days from roast date
Agtron : 70 (Whole Bean), 90(Ground)
Grinder : Nuova Simonelli Mythos One Climapro Grinder
Tamper : Pergtamp
Portafilter Basket : VST 20g
Espresso Machine : Slayer V3
Coffee Refractometer : VST Coffee III
Calibration : 20g Dose, 40g Yield, Shot : 14s Pre-infusion slow ramp to 3 bar, 16s 9 bar full pressure, total brew time: 30s
Water temperature : 93.3°C. Final EY 18.70% (tapping distribution).
Dose Tolerance : 20g ± 0.1g
Shot Tolerance : 40.0g ± 0.5g
The four methods of distribution that we will be testing is :
North South East West (NSEW) :
OCD Tool (Shout out to knockhouse supply co. for lending this to us!) :
1. Pull 10 espresso shots using the calibrated settings using the Tapping Distribution Method.
2. Pull 10 espresso shots using the calibrated settings using the NSEW Distribution Method.
3. Pull 10 espresso shots using the calibrated settings using the Stockfleth Distribution Method.
4. Pull 10 espresso shots using the calibrated settings using the OCD Tool Distribution Method.
5. Wait for espresso shots to cool to room temperature.
6. Measure TDS and EY percentage for each sample.
|Tapping Method||NSEW Method||Stockfleth Method||OCD Tool Method|
In our first round of testing, we had different baristas carry out each distribution method as they were more familiar with that particular style. However, we felt that it produced a biased and inconsistent result. Hence we carried out a second round of testing with only one barista performing all the espresso preparation steps. The results posted above are measurements from the second round of tests.
Since we set out to find the best of the distribution method rather than compare a null and alternative hypothesis, we decided to use the Tapping Distribution Method as the null hypothesis (Control Method) and compare the other distribution methods to it. For those unfamiliar with statistics, statistical significance is calculated using a p-value, which tells you the probability of your result being observed, given that a certain statement (the null hypothesis) is true. Most academic research (Disclaimer again: this is far far from any academic standards) set a minimum significance level of 0.05. This means that for a particular hypothesis that you are proposing, if it produces an insignificant result more than 5 times out of a 100 attempts, it is not statistically significant and you cannot argue that your proposed hypothesis is valid. This is to prove that it is not a result of luck or chance but rather, has an actual correlation. Hence, if the final P-value that you calculate is greater than 0.05, your proposed hypothesis is not significant enough to replace the null hypothesis (in this case the Control Method).
In conclusion, based on the P-Values that we obtained from our tests, none of the methods have proved to be statistically significant. That means that none of the methods are decisively more superior than the other! Does that mean there is no King of Distribution Methods? Not so fast, there are still many ways we can conduct an experiment that provides a better and more accurate insight into espresso distribution methods. First of all, this experiment was carried out only with a small sample size of 40 espresso shots. A bigger samples size would definitely prove to yield much better and more accurate results. As you can see, the Stockfleth method is also very close to statistical significance so a bigger sample might help to clarify the results. Another possible flaw is the inaccuracy of tamping and pre-infusion methods. There is the possibility of inconsistency of tamping which results in different tamping angles and puck density. Pre-infusion could also cause a difference in extraction. For a future follow-up study, we could possibly look at bigger samples sizes using no pre-infusion methods in brewing and using a puqpress or levy tamp which would help to elimate these inconsistencies.
Last but not least, if you like the content that we are giving out don't forget to subscribe and support us by buying some awesome coffee here!