Dr Monika Fekete investigates the role of minerals in brew water, asking whether they influence extraction or if it’s just our taste buds.

“You must realise the power that water has,” explains Yanina Ferreyra, the recently crowned Australian Specialty Coffee Association Australian Brewer’s Cup champion. Reflecting on her journey, she stresses “water plays a gigantic role, but there is much left to discover. Working with water opens up lots of questions”.

I couldn’t agree more. Experience shows that the mineral make-up of water has a substantial and even somewhat predictable effect on the taste outcomes of coffee. This immediately presents the question: what happens exactly? How do minerals modulate the taste of coffee?

I have recently reviewed the scientific literature to date on the role of minerals on taste outcomes (see R&D Lab article in BeanScene December 2018). Briefly, I found that while the role of bicarbonates is fairly well understood, the all-important part played by cations (such as magnesium, Mg, and calcium, Ca,) remains obscure. 

As proposed back in December, I set out to investigate if the key role of cations lies in aiding extraction or if they mainly act on our taste buds to affect the final taste experience.

As it turns out, taste perception has more to do with it than we might think. 

A sensory experiment

My goal was to test if a panel of coffee experts could taste a significant difference between coffee brewed with minerals added into the brew water (salts added pre-brew) as compared to the coffee sprinkled afterwards with the same salts (salts added post-brew). 

If the salts aid or block the passage of flavour compounds by chemically interacting with them during extraction, the pre-brew result should be quite different to the post-brew addition, where the salts never had the chance to work their magic on the extraction. In this case I would expect the panel to prefer one method over the other. 

If there was no preference either way, we could argue that extraction proceeds similarly with or without the added cations, and that they really only need to be present in the final brew in order to create a particular taste profile.

Experimental design

The trial had to be carried out in a blind, randomised manner to minimise unconscious bias. Judges were not informed about the goal of the experiment to ensure they had minimal information that could affect their scoring. Randomisation helped to remove potential confounding factors as I’ll explain below.

With the help of my hosts at Bennetts, we decided that 12 filter coffee samples would be close to the maximum that the panel could taste without significant palate fatigue. This allowed us to use two different coffees: a Colombian natural (Coffee 1), and a Guatemalan washed (Coffee 2). The two coffees were brewed using each method (with salts added pre- and post-brew), so we had four different samples altogether. Repeat samples were essential to account for any variation. Three repeats of the four samples quickly built up the total of 12. 

Sample preparation

All coffees were prepared in a separate room using the same automatic drip filter method: 120 grams of pre-ground and mixed coffee was brewed with two litres of water over a five-minute brew cycle into a thermos pot. The only exception was the water recipe. 

For pre-brew samples, pure water was mixed with mineral concentrates to yield a final concentration of 75 parts per million (ppm) Ca and 75 ppm Mg. These levels were chosen based on previous taste tests to represent a significant but not completely overpowering difference compared to tap water. 

The post-brew samples were prepared without any Ca or Mg added into the water. The minerals were added to the brewed coffee to result in the same final concentration in the coffee as in the pre-brew samples, taking water retention in the spent grinds into account.

Tasting and scoring

The panel of 13 judges was carefully calibrated.

Samples were then served to the tasting panel in identical cups. The order of sample allocation to tasting station was randomised to make sure no one would always taste the first pump or the bottom of the jug. 

The panel was asked to assign scores ranging five to 10 (average to exceptional) in 0.5-point increments for acidity, sweetness, body, balance, aftertaste, and overall attributes. They were also asked to list main tasting notes, without any lists or prompts. 

In one sitting, all four samples were tasted once in a randomised order. The panel then had a short break, before returning to another two sittings to taste the same samples again in a new, random order. The samples were featured equally in each sitting to account for the progressive fatigue of the panel. 

To me, crunching the numbers was the most exciting part of the experiment. The time has finally come to deliver the verdict. 

The result was very clear. The statistical test (non-parametric ANOVA) showed two main conclusions.

First, the panel had a strong preference for Coffee 1 over Coffee 2. As shown in Figure 2, Coffee 1 (red and orange) was given a median score of 7.5 for the overall attribute, a full point higher than Coffee 2 (purple and blue). (The median score is shown as the thicker line inside each box. Inside the box are commonly given scores, while the whiskers show the extremes.) The same trend was observed for all other attributes.

This result tells me that the scores given by the panel well reflected their preferences. They weren’t scoring at random. This makes the second observation even more interesting. 

There was virtually no difference in the scores with respect to the brew method used. 

For the overall attribute, the median score was equal for pre-brew and post-brew salt addition in the case of both Coffee 1 and Coffee 2. There was no statistically significant preference for either brew method for any sensory attribute. 

Tasting notes were given freely, and frankly, I didn’t have great expectations of their usefulness. The result left me surprised. 

Look at the word clouds over each sample. Larger fonts represent the most common tasting notes, smaller fonts the less common ones. The key notes for Coffee 1/ pre-brew (red) were plum, lemon, juicy, winey, chocolate, sweet, apple, and orange. Some caramel and a hint of dryness. These are nicely echoed in Coffee 1/post-brew (orange): winey, chocolate, plum, dry, apple, caramel, bitter, with some sweetness, orange, and juicy notes. Coffee 2 was less liked, which is reflected in notes such as dry, woody, bitter, smoky, burnt, with caramel as a redeeming feature. The similarity between pre-brew (purple) and post-brew (blue) is glaring.

Like any experiment, this one has its limitations. However, I would argue that the results present a strong case to challenge our understanding of how minerals affect the taste of coffee. 

Adding calcium and magnesium salts to coffee that was brewed with almost perfectly pure water yielded a taste experience indistinguishable to another coffee brewed with a high level of added minerals.

Is it all in our tastebuds then? Well, they might just play a bigger part than we expected.

This article appears in FULL in the April 2019 edition of BeanScene Magazine. Subscribe HERE.