Understanding Whisky Production Unit 7: Sensory Assessment & Microbiology
Chapter 16: Sensory Assessment: Tasting Panels, Triangle Tests, and Flavour Wheels
Preparing spirit samples at 20% ABV, controlling sensory bias, difference testing (triangle, duo-trio), and descriptive flavour profiling.
On this page
Gas chromatography can identify every individual congener in a spirit sample down to parts per billion, but only human sensory panels can evaluate how those molecules harmonise on the palate.
Professional nosing requires careful control of sensory bias. Samples are uniformly diluted to 20% ABV with demineralised water to dampen the anaesthetising burn of concentrated ethanol and liberate volatile aromatic esters. Tasters work in neutral, odour-free booths with blinded, randomised sample codes, using formal discrimination protocols like triangle testing to confirm whether new process tweaks have altered the underlying character of the spirit.
Controlling Flavour and Appearance
In this section, we will consider the importance of controlling the flavour and appearance of spirits. To do this, we must understand the types of compound that impart these sensory characteristics and consider where they come from.
The Importance of Controlling Flavour and Appearance
The sensations that we perceive when consuming spirits, namely flavour and appearance, are central to its success in the market place. A brand only succeeds if it has the sensory characteristics that the consumer is looking for. Desirable character varies widely from one spirit drink to another. For example, we expect very different characteristics from vodka compared to a well-aged cognac.
The Types of Compounds that give Flavour and Appearance
The four groups of sensory characteristics, summarised in the table at the start of this lesson, result from the presence of different compounds. These compounds are:
- Aroma: The volatile compounds that are released from the liquid into the surrounding air give the spirit its aroma. You do not have to consume the product to detect these compounds. We will return to this point later in this chapter. There is a diverse range of volatile compounds, with varied chemical structures, that gives rise to a wide spectrum of different aromas.
- Taste: Taste characteristics are simpler. There are five primary tastes: bitter, sweet, sour, salty and umami (savoury taste). The first three tastes are most relevant to spirits. A limited number of compounds impart these tastes. Bitterness usually comes from tannins, sweetness from sugars and sourness from acids.
- Texture: Currently, we have a relatively poor understanding of the compounds that give texture to a spirit. However, we do know that the alcohol in the spirit majorly contributes to mouthfeel. (Think about the warming sensations that arise when consuming a spirit drink.)
- Appearance: Compounds that both give colour, and affect the spirit’s clarity, will influence its appearance.
Where do these Flavour Compounds Come From?
Flavour compounds can come from the raw materials. For example, the malty aromas in a whisky, or juniper and citrus notes from the botanicals used to make gin. However, the production process also influences flavour, with fermentation being a key contributing stage. Yeast, as well as producing alcohol, also produces a large number of flavour compounds. Flavour is then further modified by distillation and, where applied, maturation.
Although not permitted for all spirits categories, the simplest way to modify and control flavour is to add flavourings (natural or chemical food flavours). Simply add more or less, or different flavours. In spirits categories where flavourings are not used, such as Scotch Whisky, control relies on knowing where flavour compounds come from in production. Once you understand this, you can run the process in a way that gives your desired flavour.
Vodka and Neutral Spirit
In this section, we have explored the compounds that give flavour. Let us now turn our attention to vodka and neutral spirits. In this case, less is more! The aim with both of these spirits is to remove unwanted flavours. This will give the cleanest spirit possible. To do this, you still need to understand the types of flavour that can be present, and where they come from. This will then help us to remove them.
Using Sensory Assessment to Control Flavour
We will now look at the options available to spirits producers, to help them measure and control flavour. We will also consider why sensory assessment is generally the preferred option.
In the previous section, we learned why it is important to control flavour. So how exactly can we control flavour?
In an ideal world, our distilleries and bottling halls would have instruments that continuously measure flavour. These instruments would alert us if the flavour started to drift out of spec. However, such machines do not exist. The human nose is much more sensitive than any current instrumental measures. It detects flavours that would not register on instruments. Therefore, the industry has to rely on sensory assessment, which is where human assessors evaluate the products.
We should note here that we could measure appearance (colour and clarity) using instruments. For the remainder of this lesson, we will instead focus on sensory analysis as a mode for measuring flavour.
Sensory Analysis
What does carrying out a sensory assessment aim to achieve, and how are we going to achieve it?
Aims of Sensory Analysis
As mentioned previously, the ideal situation would be to have an instrument that measures flavour. Instead, we rely on the power of human sensory experience. We must, however, make this assessment as similar/accurate as an instrumental measurement as possible. We must avoid any factors that might influence or alter the information collected from the assessors.
Sensory analysis should be carried out under carefully controlled conditions. It should be unbiased and free from personal preference. We want to measure flavour, not find out how much an individual likes the product. This is a very different approach from consumer testing.
The Sensory Assessor
We need to decide who is going to carry out the sensory assessment. There are generally two options:
- The single sensory expert - typically a master blender
- A group of trained assessors - a sensory panel
There are pros and cons with each option. Companies frequently use a combination of the two.
A master blender is an individual with an intimate knowledge of the flavour of their product/brand. They undergo years of training and have a wealth of experience. The advantage of having a master blender is that, not only do they pick-up problems or drifts in product quality, but they also have the experience to know what might have caused these issues. Furthermore, they can usually suggest how we can correct these problems. The master blender’s availability can, however, be a problem - what happens when they are sick or called away on business?
The alternative is to use a sensory panel. This is a group of individuals, who are specifically trained to carry out sensory assessments. They do not have the same degree of expertise or specialised product knowledge as the master blender, but there are some advantages in using them. Firstly, their availability is better. It is unlikely that everyone on the panel will be unavailable at the same time. The other positive is that they represent a wide range of individual sensitivities. We all have strengths and weaknesses in our ability to detect different flavours. It is also easier to recruit a new panellist, and train them to the required level, than it is to replace a master blender. With the latter, succession planning is essential.
Master blenders and panellists have to follow certain rules. These rules include the following:
- Do not wear perfume or aftershave
- Do not use strong-smelling hand creams
- Do not eat prior to assessing samples
- Do not smoke prior to assessing samples
Preparing Spirit Samples for Sensory Evaluation
A sample must be prepared and presented in a way that allows easy assessment of its sensory characteristics. The way in which the sample is prepared depends on the type of assessment being used.
Due to the high alcohol content of spirits, most of the routine sensory assessments are based on aroma only. It is often unfeasible to carry out taste tests on large numbers of samples. This is due to health and safety reasons, and also because the high level of alcohol limits the number of samples that can be assessed in one sitting, due to sensory fatigue.
We refer to the assessment of a spirit’s aroma as nosing. For nosing, samples are presented in glasses with a small aperture and are covered with a lid or watch glass. This allows the volatile aromas to build up in the glass.
Coloured glasses can be used to disguise the appearance of the sample, as this can influence perceptions. For example, if a cask-matured sample is dark coloured, the assessor will expect it to have more mature flavours.
When nosing, samples are typically diluted with water, to around 20% abv. With certain spirits, this encourages the release of aroma compounds from the liquid into the headspace above. Dilution also reduces the burning sensation in the nose associated with alcohol, making the other flavours easier to detect. It also allows us to assess more samples in a single session, as there is less sensory fatigue and less alcohol burn.
If we want to evaluate the taste or mouthfeel characteristics of a sample, then we must use a taste test. The samples can again be diluted, though this is often not the case as alcohol itself is an important aspect of taste and mouthfeel.
As mentioned earlier, it is important that the sensory assessor does not have any preconceived ideas about the samples. To overcome this, samples are coded before being presented to the panel. Three-figure random codes are often used (e.g. 587, 312). You might think that coding your samples A, B, C or 1, 2, 3 would be simpler, but even this can influence perceptions. The assessor might assume that A is superior to B, or that 1 is superior to 2 - which is, of course, not necessarily correct.
Sensory Test Room Conditions
The ideal test room is one where the sensory panellist can carry out their own assessments without the influence of others. Individual booths are best for this. The room should be free from any interfering aromas. It should also be at a comfortable temperature, with controlled lighting. To disguise the appearance of the sample, we could use coloured lights as an alternative to coloured glasses. Although this is the ideal, it might not be practical in a distillery or bottling hall environment. In this case, as many factors as possible should be controlled.
The most important factor to control in the sensory room is to ensure that background aromas are minimal.
Types of Sensory Tests
In the next part of this lesson, we will look at different types of sensory tests and the information that they provide.
Selection of Sensory Tests
Sensory analysis is a time consuming and costly process. It is thus important to select the right type of sensory test. The best test is one that will provide all of the information required, without collecting additional, unnecessary data. Sensory analysis can be divided into three main types of test:
- Sample screening
- Difference testing
- Descriptive analysis and flavour profiling
We will now look at each of these in more detail.
Sample Screening
Sample screening is the simplest form of sensory analysis.
It involves the assessment of a sample, or more commonly a set of samples, to determine whether they have the desired flavour characteristics. We can use a pass/fail scoring system, or we can grade the samples in terms of how close they are to the desired character. When using a less experienced panel, we can give them a reference sample that represents the target flavour. The panel then compares the test samples against the reference samples. More experienced panels, or master blenders, will not require a reference. They will know, through experience, the type of flavour that is required. We can use these screening tests at different stages in the production process, anywhere from an evaluation of the raw materials, to a post-bottling quality check.
Although these tests are a good, quick check of flavour quality, they do not provide detailed information on the nature or degree of any difference.
Difference Testing
As the name suggests, we use these tests to determine whether there is a difference between samples.
They are carefully controlled tests, carried out using approved sensory methods as defined by the International Standards Organization (ISO standards). You might be asking: “Is it necessary to take a formal approach?” In order to answer this, let us take a few minutes of reflection.
Key Operating Insight
Welcome back. So what were your thoughts? Maybe you fully trust the sensory abilities of your colleagues and are confident that they will come back with the correct answer. However, maybe you have realised that there are potential factors that could interfere with the feedback you receive. The colleagues’ personalities could be a determining, and in turn misleading, factor. You might have selected someone who is keen to prove that their senses are more acute than others are. Therefore, they insist that they can pick up differences between the samples that you know are actually the same. At the other end of the personality spectrum, is the individual who lacks the confidence to say that they can detect a difference, even when it is obvious. These are the problems associated with individual assessments. A group approach might be better. However, there might be one dominant individual, who will influence the overall decision. You can overcome all of these issues by using a formal, unbiased test. The next section explains how.
There are several types of sensory difference tests available; the most commonly used is the triangle test. A triangle test determines whether there is a difference between two samples. Panellists are presented with three coded samples. Two of these samples are the same, and one is different.
We ask the panellists to pick out the odd sample. If they cannot detect a difference, they are asked to guess. This is referred to as a forced choice test. It overcomes the problem that some people may detect a difference but are not confident enough to say so. Each panellist carries out the test individually, overcoming the influence of a dominant individual. In this test, we can say that the samples are different if the number of correct responses is greater than the number that would be obtained from guesswork alone. We can use statistical tables to interpret the tests. We can find these tables in the relevant ISO standard, or online.
Other types of difference tests are available. What they all have in common is that they are forced choice tests, which are interpreted based on the number of correct responses relative to the number that would be obtained by guessing.
The tetrad test is a four-sample test, whereby a panellist is asked to match up two pairs of samples.
Another common difference test is the duo-trio test. Here, we give panellists a reference sample and ask them to match one of two possible test samples with this reference.
Difference tests tell you whether samples are different. However, they do not provide any information on the nature of these differences. For this, you need to carry out a descriptive test, which we will cover in the next section.
Descriptive Analysis and Flavour Profiling
Descriptive analysis is a general term used for any assessment that involves producing a description of a sample’s sensory characteristics.
We commonly use flavour descriptions when marketing spirits. These descriptions aim to entice the consumer to purchase the product. Consequently, they often use flowery language. However, descriptive analysis used during the manufacture of the product is different, as the aim is to measure and control flavour. The results of this sensory assessment need to be communicated from the master blender, or sensory panel, back to production. It is thus important to use well-defined and easily understood words. In the 1970s, the Scotch whisky industry realised the need for a common language to aid communication. This led to the Scotch Whisky Research Institute flavour wheel.
This wheel is widely used in the industry, and similar wheels have been developed for other spirit categories.
The wheel covers all of the different flavours that we need to measure and control. This includes both desirable flavours, such as fruity and peaty, but also includes off-notes, such as sour and stale. It is a production tool, not a marketing tool. It optimises desirable flavour and controls negative attributes. It covers all aspects of flavour, encompassing aroma, taste and mouthfeel.
Flavour wheels give us a common language with which to describe flavour. Using the flavour wheel above, we might describe a whisky as having fruity, malty and vanilla notes. However, this does not tell us anything about how strong these flavours are relative to one another.
Flavour profiling is the next step up from these simple descriptions. This involves scoring the intensity of each flavour. We give the sensory panel a list of flavour attributes that they are required to score. The figure below shows an example of a scorecard for gin evaluation. We can collect scores using paper ballots. However today, we tend to use sensory data collection software.
Once the panel has completed this exercise, the average score is calculated for each flavour. The scores are then plotted to produce a flavour profile. We typically use radar plots for this. We often call these profiles a spider web diagram. The figure below exemplifies a flavour profile of a gin:
Flavour profiles quickly and easily show which flavours are most dominant and help to identify any off-notes. From the flavour profile above, you can see that this gin has juniper, green and spicy aromas, and is free from off-notes such as sulphury, sour and musty. We can compare two or more samples by plotting them on the same axes. Flavour profiling thus provides the most comprehensive way of measuring flavour and comparing samples using sensory analysis.
When to Carry out Sensory Assessments
In the final part of this lesson, we will consider when we should carry out process sensory assessments.
All stages in production, from the raw materials through to bottling halls processes, have the potential to influence flavour. We must continually monitor flavour. In the early production stages, screening tests are often used to provide a rapid quality check. For example, for matured products, the spirit is checked before it is put into the cask. This identifies problem samples (off-notes). If off-notes occur, we can change the process or, in the worst instance, halt production.
We can employ difference tests if we suspect there is likely to be a flavour change, e.g. in a less routine application. One example might be if we have switched to using a new type of yeast. We could use a difference test to compare the spirits that were produced using the two types of yeast (old and new) to determine if the change has had a detectable effect. We can use difference tests to compare samples from batch-to-batch, or year-to-year, in order to check flavour consistency. Difference tests are also frequently used to check customer complaint samples, by comparing suspect samples with library/ reference samples.
Descriptive analysis, particularly flavour profiling, is the most time consuming (and therefore expensive) sensory approach. However, it provides detailed information on flavour character and flavour differences. We tend to use this type of test more for evaluation of final product, or during product development. We can also use it to gain a detailed understanding of any flavour changes that have occurred.
Key Takeaways
Dilution to 20% ABV is essential for sensory assessment, breaking the ethanol-water matrix to release volatile aroma compounds without fatiguing human olfactory receptors.
About the author
