Winemaking Laboratory
Sensory science and tasting / Systematic tasting

Chapter VII · Sensory science and tasting

Systematic tasting

Systematic tasting turns the assessment of wine into a consistent, repeatable procedure: colour, aroma, taste, harmony. At international competitions wine is scored on the OIV 100-point system, and sensory tests and statistics are used to prove a difference between two wines.

Score (still wine)

Visual

Nose

Taste

Overall

The OIV standard for international competitions (still wine): each criterion is scored on five levels, with points weighted by the importance of the criterion; the maximum is 100. Medals: Grand Gold ≥ 92, Gold ≥ 85, Silver ≥ 82. The number of medals at a competition is limited (no more than 30% of samples in total). For sparkling wine the sheet adds a criterion for foam and bubbles.

Total—points
Result—
Nose—points
Taste—points

Test result

In a triangle test each taster receives three samples, two of them identical, and must pick the odd one. The chance of guessing correctly is 1/3; in the duo-trio and paired tests it is 1/2. The null hypothesis: the wines do not differ. The p-value from the binomial distribution: P(X ≥ x) assuming everyone is only guessing. If p < α, the difference is statistically confirmed (ISO 4120, ISO 10399, ISO 5495).

Correct answers—%
p-value—
Conclusion—
Minimum needed—

Planning a test

If a share pd of tasters truly perceives the difference and the rest guess, the probability of a correct answer is p = pd + (1 − pd)·p₀. The required panel is the smallest n for which an exact binomial test (α) detects this difference with probability ≥ 1 − β. A typical guide: pd = 30%, α = 0.05, β = 0.2. Detecting a small difference needs a large panel.

Tasters needed—
Critical number—
Power—%
Actual α—

Systematic tasting

Tasting is a tool for assessing wine. For the result to be repeatable and comparable, it is always carried out in the same order:

1. Visual assessment: limpidity (clear, hazy, cloudy), colour and its intensity (tilting the glass against a white background), the rim hue (an indicator of age), and in sparkling wine the bubbles.

2. Nose: first the still glass (the most volatile compounds), then after swirling. Assessed: cleanness (absence of faults), intensity, character (fruit, flowers, plants, wood, ageing notes) and development.

3. Palate: a small sip held in the mouth for 10–20 seconds, drawing in a little air (retronasal aroma). Assessed: sweetness, acidity, tannin, alcohol, body, flavour intensity and length (how many seconds the aroma remains after swallowing).

4. Overall assessment: balance, complexity, typicity, ageing potential.

Tasting conditions

• Glass: tulip-shaped, clear, the ISO 3591 standard tasting glass (≈ 215 mL), filled to a half or a third.

• Room: well lit (daylight), free of odours, at 18–22 °C.

• Wine temperature: according to type (see "Aroma compounds").

• Order: dry before sweet, white before red, light before full, young before old.

• Blind tasting: samples are coded so that the label and price do not influence the assessment.

The OIV 100-point system

The International Organisation of Vine and Wine (OIV) uses a single score sheet for international competitions. For still wine:

SectionCriterionExcellentVery goodGoodAveragePoor
VisualLimpidity54321
VisualAspect108642
NoseGenuineness65432
NosePositive intensity87642
NoseQuality161412108
TasteGenuineness65432
TastePositive intensity87642
TasteHarmonious persistence87654
TasteQuality2219161310
HarmonyOverall judgement1110987

The points are weighted by the importance of each criterion: nose and taste together carry 74 points. Medals: Grand Gold ≥ 92, Gold ≥ 85, Silver ≥ 82. The jury result is the median of the experts' scores.

Other common systems are the UC Davis 20-point scale (traditionally for teaching and technical assessment) and the journalists' 100-point scale, which in practice starts at 50.

Difference tests

When you need to know whether two wines differ (for example, made with different yeasts or closures), an expert's opinion is not enough. Sensory tests and statistics are used.

• Triangle test (ISO 4120): the taster receives three samples, two of them identical, and must find the odd one. The probability of guessing correctly is $p_0 = 1/3$.

• Duo-trio (ISO 10399): a reference and two samples; which matches the reference? $p_0 = 1/2$.

• Paired directional test (ISO 5495): "which is more acidic / sweet / astringent?" $p_0 = 1/2$.

Under the null hypothesis the wines do not differ and everyone is only guessing; the number of correct answers then follows the binomial distribution:

$$P(X \geq x) = \sum_{k=x}^{n} \binom{n}{k} p_0^k (1 - p_0)^{n-k}.$$

If this probability (the p-value) is below the significance level $\alpha$, the difference is confirmed. For example, in a triangle test 13 correct answers out of 24 give $p \approx 0.03$.

Panel size

If a share $p_d$ of tasters truly perceives the difference, the probability of a correct answer is

$$p_1 = p_d + (1 - p_d)\,p_0.$$

The power of the test ($1 - \beta$) is the probability that the test detects this difference. The required panel size rises quickly as $p_d$ falls: in a triangle test at $\alpha = 0.05$ and $\beta = 0.2$, about 40 tasters are needed for $p_d = 30\%$ and about 87 for $p_d = 20\%$.

Sources

OIV, "OIV Standard for International Wine and Spirituous Beverages of Vitivinicultural Origin Competitions" (OIV-CONC 332A/2009); ISO 4120:2021 (triangle test), ISO 10399:2017 (duo-trio), ISO 5495:2005 (paired comparison), ISO 3591:1977 (tasting glass); H. T. Lawless, H. Heymann, "Sensory Evaluation of Food: Principles and Practices", 2nd ed. (2010); R. S. Jackson, "Wine Tasting: A Professional Handbook" (2017).

How to use this page

1. In the "OIV score sheet" mode, choose a level for each criterion and the points are calculated automatically. The cards show the total, the medal and the nose and taste scores; the chart shows each section as a percentage of its maximum.

2. In the "Triangle test" mode, choose the test type and enter the number of tasters, the number of correct answers and the significance level. You will see the p-value, the conclusion, the estimated share of discriminators and the minimum number of correct answers. The chart shows the distribution under guessing, with the critical region in red.

3. In the "Panel size" mode, choose the test, the share of discriminators you want to detect, α and β. You will see how many tasters are needed and how many correct answers will prove the difference.

Practical tips

• In a triangle test use all six sample orders (AAB, ABA, BAA, BBA, BAB, ABB) equally and label the glasses with random three-digit codes.

• Tasters must always give an answer, even if they guess ("forced choice"): the statistics are built on this.

• Samples should be at the same temperature and volume; if colour is not the subject of the test, hide colour differences with dark glasses or red light.

• Do not taste more than 6–10 samples in one session: taste receptors tire. Rinse the mouth with water between samples.

• Always keep tasting notes in the same format: over time they become the best record of how your own wine evolves.

1. In a triangle test, 13 of 24 tasters found the odd sample. What does this mean?

With pure guessing 24/3 = 8 correct answers are expected. P(X ≥ 13) from the binomial distribution ≈ 0.03 < 0.05, so the difference is statistically confirmed.

2. Why is taste "quality" worth 22 points on the OIV sheet, but limpidity only 5?

The OIV sheet gives each criterion different points according to its importance: nose and taste together carry 74 points, the visual 15 and overall harmony 11.

3. Why is a large panel needed to detect a small difference?

The power of a test depends on the gap between the expected and chance proportions and on the sample size. To detect pd = 20% in a triangle test at α = 0.05 and β = 0.2, about 85–90 tasters are needed; for pd = 30%, about 40.