Sensostat Try the demo

Textbook · Module 03

Sensory Laboratory, Sample Preparation and Serving

Facilities per ISO 8589, sample preparation, three-digit blind codes, serving-order balancing (Williams Latin square), and the serving plan generator.

Module 03 of 11~20 min readinteractive + quiz

Facilities for sensory testing: ISO 8589

The tasting room is part of the measuring system. ISO 8589:2007 (+Amd 1:2014; Russian counterpart GOST R ISO 8589-2010) sets requirements with a single goal — to remove everything that could bias the score: odours, sounds, visual cues, communication between assessors.

Test area

Individual booths with partitions, neutral lighting at 70–80° (and red light for masking colour), extract ventilation with positive pressure, silence, temperature 20–22 °C.

Preparation area

A kitchen/laboratory next to the booths, separated so that the odours and sight of preparation do not reach the assessors; a sample pass-through window into each booth.

Office area

The panel leader's workplace, data storage, results analysis. Assessors walk to the booths bypassing the preparation area.

Layout of a sensory laboratory per ISO 8589 (hover over a zone)

Test area: 6 booths B1 B2 B3 B4 B5 B6 sample pass-through windows from the preparation area ↑ tablet/PC in each booth · neutral light · extraction Sample preparation area scales · cooker · fridge · code labelling · pass-through window above Panel leader's office PC with data-collection software protocol archive Waiting area assessors enter from here — they do not pass the preparation area
The key layout principle: the assessor neither sees nor hears sample preparation or other assessors. Small enterprises may use collapsible booths — what matters is not capital renovation but the control of odours, light and communication.

Sample preparation: the "identical for everyone" rule

  • Homogeneity: all portions from one batch; mixing, identical cutting, identical portion weight (e.g. 10–15 g of chocolate, 30–50 ml of a drink).
  • Serving temperature is a critical parameter: drinks at 10–12 °C or room temperature, hot dishes at 60–65 °C, chocolate at 20–22 °C. A difference of 3–5 °C noticeably changes the perception of aroma and viscosity.
  • Tableware: identical, neutral, odourless (glass, porcelain, food-grade disposable plastic); opaque or coloured for masking colour.
  • The preparation → serving time is fixed: bread is evaluated at an equal time after baking, coffee at an equal time after brewing.
  • Sample load: 6–8 samples per session for taste tests; more for olfactory tests, but with breaks. Strong/persistent tastes (pepper, alcohol) shorten the series.
  • Palate cleansers: drinking water at room temperature — always; unsalted crackers or bread — for fatty/spicy products; a mandatory 30–60 s pause between samples.

Blind coding

Each sample gets a random three-digit code (e.g. 483). Code requirements:

  • no "meaningful" coincidences with reality (not 001/002 — order; not 100 vs 200 — a hint at versions);
  • no emotionally charged or memorable combinations (777, 666);
  • different samples in one serving get different codes; the same product gets different codes for different assessors;
  • codes are generated from a pool with a fixed RNG seed, so that the serving plan is reproducible and auditable.

Balancing the serving order

The order in which an assessor receives samples itself affects the scores (position effects, contrast). That is why the serving design is balanced:

  • Triangle test — 6 possible triads: AAB, ABA, BAA, BBA, BAB, ABB. They are distributed evenly among assessors; the order within a triad is randomized by the plan.
  • Several samples — Williams Latin square: every product appears at every position and follows every other product an equal number of times — balanced both for position and for carry-over effects.
  • More samples than a session can handle — balanced incomplete block designs (BIB, ISO 29842): each assessor sees a subset, but product pairs occur evenly.
  • Consumer tests — the standard is sequential monadic serving: one product at a time with a full questionnaire for each.

Serving plan generator

A session plan in a second interactive

This is how the "serving plan" works in Sensostat: choose a method and panel size to get a balanced design with three-digit blind codes. The RNG seed is recorded in the protocol; the plan is reproducible and auditable.

Press "Generate the plan".

What the assessor questionnaire looks like

A minimal triangle-test questionnaire contains: identifiers (assessor, session), the codes served, one mandatory forced-choice question and an optional comment. Try it — the mock-up is functional:

Triangle test · Session 14
Assessor A-07 · Samples served left to right · ISO 4120

Before you are three samples. Two of them are identical, one is different. Taste the samples from left to right and mark the odd one. A "don't know" answer is not allowed — choose the most likely one.

Why there is no "no difference" button: discrimination tests are forced-choice procedures. An "I cannot tell" answer would break the binomial model: the guessing probability p₀ = 1/3 assumes that every assessor picks one sample out of three.

Further reading

Test yourself

module 03 quiz

Three questions on test organization

1. Why do taste tests use red lighting or opaque tableware?

Colour shapes expectations ("dark = bitter"). When taste is being tested, the visual channel is switched out of the equation.

2. A Williams Latin square is chosen over an ordinary Latin square because it…

The Williams square balances first-order carry-over effects: every product follows every other product an equal number of times.

3. Sample codes 100, 200, 300 for three recipes are…

Codes must be random three-digit numbers with no meaning. 100/200/300 involuntarily rank the samples in the assessor's eyes.

The infrastructure is ready — on to the methods

Module 04 is the heart of analytical sensory science: discrimination tests with a live statistical calculator.

Module 04 →