Can you taste blending water? The sensory check no certificate of analysis will do for you

Key takeaways

• The nose beats the instrument at the concentrations that matter. A trained water panel detected geosmin at 2.5 ng/L and 2-methylisoborneol at 4 ng/L, because “the odours and flavours are produced by substances that can be in very low concentrations that, although perceptible by humans, cannot be measured by instrumental methods” (Water / SUSPAW study).
• Taste is sharper than smell for the worst offender. WHO gives the taste threshold for 2-chlorophenol as 0.1 µg/L against an odour threshold of 10 µg/L — a hundredfold gap (WHO, Guidelines for Drinking-water Quality).
• The fault travels into the bottle. Geosmin “can survive every step of the spirits production, from fermentation, to maturation,” which is why “much care is needed in choosing water source, especially that for spirit reduction” (Lallemand Biofuels & Distilled Spirits).
• Sensory testing is written into the standards, not bolted on. India’s IS 10500 opens its table with odour and taste, both required to be “Agreeable”, tested “cold and when heated” and “at several dilutions” (IS 10500:2012). Europe has a dedicated method, EN 1622, for putting a number on it.
• For blending water the pass criterion inverts. Drinking water is judged on being pleasant; WHO’s own review notes water at 25–50 mg/L TDS was “described tasteless” (Kozisek, in WHO, Nutrients in Drinking-water). For spirit dilution, tasteless is the specification — so the panel is testing for absence, not for quality.

Yes — and at concentrations far below anything a routine water certificate reports. A blending water can return a flawless certificate of analysis — conductivity under 1 µS/cm, TDS in single figures, pH in spec, zero coliforms — and still carry an earthy or antiseptic note that a trained nose finds in seconds. The compounds responsible are active at nanograms per litre, which is why the drinking-water industry never abandoned sensory testing in favour of instruments, and why any distillery buying water by specification alone is checking a third of the problem.

Why can’t a chemical spec sheet catch an off-note?

Because the analytes that ruin a water are not on the panel, and they act at concentrations three to six orders of magnitude below the ones that are.

The two canonical culprits are geosmin and 2-methylisoborneol (2-MIB), metabolites of cyanobacteria and actinomycetes. WHO puts their taste thresholds in drinking water at “a few nanograms per litre” (WHO). A Spanish study that built a water panel from scratch measured it directly: in its final training sessions, 76.7% of assessors detected geosmin at 2.5 ng/L and 77.4% detected 2-MIB at 4 ng/L, with calculated panel thresholds of 3.9 ng/L and 6.4 ng/L respectively (SUSPAW).

Set that against the units a spec sheet uses. TDS is reported in milligrams per litre; geosmin matters at nanograms per litre — a millionfold difference in scale. A demineralisation train can strip every ion it was designed to strip and leave a neutral organic molecule untouched, and the water will still read as pure on every number you asked for. The study puts the reason for keeping people in the loop plainly: “The odours and flavours are produced by substances that can be in very low concentrations that, although perceptible by humans, cannot be measured by instrumental methods.” For this class of compound, human olfaction is not a fallback for the lab — it is the more sensitive detector.

The second family is worse, because we make it ourselves. Chlorophenols form when chlorine or chloramine meets phenolic organics in water, and their thresholds are startling:

Source: WHO, Guidelines for Drinking-water Quality — acceptability aspects

Two things follow. First, taste is up to a hundred times more sensitive than smell for these compounds — the one case where nosing alone will miss what the mouth catches. Second, look at the compliance limit alongside them: IS 10500 caps phenolic compounds (as C₆H₅OH) at 0.001 mg/L, i.e. 1 µg/L acceptable, 0.002 mg/L permissible (IS 10500:2012). That limit is on the parent phenols. Their chlorinated derivatives are detectable by taste an order of magnitude lower. A water can be fully compliant on paper and taste of a hospital.

What does an off-note in the water actually do to the spirit?

It becomes an off-note in the spirit, in proportion to how much water is in the bottle — which is a great deal more than most people assume.

Water is not a trace ingredient in a bottled spirit. Reducing a cask at 63.5% ABV to a 46% bottling makes roughly 28% of the finished liquid added water; taking a 96% ABV neutral spirit down to a 40% vodka makes it about 58% (approximate volume fractions — the exact figures require the contraction correction covered in our guide to the proofing arithmetic). Anything carried in that water is carried into the glass at close to full strength.

And it does not get filtered out along the way. Lallemand’s technical note on spirit aromas is unambiguous about geosmin: it comes from “two main sources: water and raw material,” and “it can survive every step of the spirits production, from fermentation, to maturation, thus appearing in the final product.” Their conclusion is written for exactly this decision — “the fact that water is used in many parts of the process means that much care is needed in choosing water source, especially that for spirit reduction” (Spirits Selection / Lallemand).

The chlorophenol risk is sharper still for whisky, because whisky is a phenol-rich matrix by design. Peated malt is prized for its phenols; introduce a chlorinated water and you are supplying both reagents. The sensory result is catalogued in the trade’s own flavour standards — 2,6-dichlorophenol is sold as a training compound described as “mouthwash,” “chemical,” “antiseptic,” “hospital-like” (FlavorActiV) — and it is the reason dechlorination is non-negotiable before water meets spirit, as covered in our comparison of tap, filtered, mineral and demineralised water.

Do the standards actually require someone to smell the water?

They do, and in India’s case sensory parameters come first in the table, before any chemistry.

IS 10500:2012 lists odour and taste among its organoleptic and physical parameters, with the requirement for both stated as “Agreeable” at the acceptable limit and “Agreeable” again at the permissible limit — one of the few rows in the standard with no numerical relaxation available. The method columns point at IS 3025 Part 5 for odour and Parts 7 and 8 for taste, and the remarks column carries the practical instructions: odour is to be tested “cold and when heated” and “at several dilutions,” while for taste the test “is to be conducted only after safety has been established” (IS 10500:2012). Those three notes are a working protocol in miniature, and they still govern the Indian regulatory floor for water used in spirits.

Europe formalises the same judgement into a measurement. EN 1622 — “Water quality — Determination of the threshold odour number (TON) and threshold flavour number (TFN)” — specifies quantitative short and full methods plus a qualitative screen, with guidance on test environment, panel selection and safety, and applies to drinking water, distribution water, raw water and migration waters from materials in contact with water (CEN, EN 1622:2006).

North America’s parallel is Flavor Profile Analysis, Standard Method 2170. FPA uses a group of four or five trained panellists to characterise a sample’s taste and odour attributes and rate their intensity, without dilution, against reference materials that give the panel a shared vocabulary (Standard Methods 2170).

How is an odour threshold actually measured?

By dilution — you keep diluting the sample with odour-free reference water until nobody can tell it from the reference, and the dilution ratio at which that happens is the number.

That ratio is the threshold odour number (TON); the same procedure judged by mouth gives the threshold flavour number (TFN). EN 1622 offers a short method — used when a sample has no odour and flavour at all, or when it only needs comparing against a specified threshold — and a full method for determining the threshold number itself, each in unforced-choice and forced-choice variants (EN 1622:2006).

For blending water the short method is the relevant one, and the target answer is the trivial one: no perceptible odour or flavour in the undiluted sample — TON and TFN at 1. FPA then earns its place as the diagnostic step, because an intensity-rated descriptor (“earthy,” “musty,” “medicinal,” “plastic”) tells you where in the plant to look far faster than re-running the chemistry does.

Who is allowed to be on the panel?

Not whoever is free that afternoon. A water sensory panel is a screened and trained instrument, and the literature is candid about how much of the recruitment washes out.

The SUSPAW team recruited 91 candidates to build a 20-person panel, recommending over-recruitment to “at least four or five times the number of assessors required” against the two-to-three-times ISO 8586 suggests. Pre-selection excluded smokers and candidates over 60; selection ran eight tests, each performed twice across four 90-minute sessions, with candidates needing 60% correct answers in at least six of the eight. Training took three phases over seven months and 22 sessions (SUSPAW). The supporting standards are a stack rather than one document: ISO 8586 for selecting assessors, ISO 8589 for booth design, ISO 3972 for taste sensitivity, ISO 5496 for odour training, ISO 5495 for paired comparison.

One detail deserves attention from anyone writing a protocol: choosing the reference water was itself an experiment, settled by triangular tests across three mineral waters with 34 assessors over 68 evaluations. Your reference water is part of your method — pick it badly and every threshold you measure moves with it.

But blending water is meant to taste of nothing — how do you evaluate nothing?

By flipping the pass criterion. In drinking water, “flat” is a defect. In blending water, flat is the specification — so the panel is not asked whether the water is pleasant, only whether anything is there at all.

The drinking-water world’s own numbers make the contrast explicit. WHO considers palatability “good” below about 600 mg/L TDS and “significantly and increasingly unpalatable” above about 1000 mg/L (WHO). Going the other way, WHO’s review of demineralised water records that water at 25–50 mg/L TDS was “described tasteless,” that the 1980 expert team recommended a minimum of 100 mg/L TDS in drinking water with an optimum of 200–400 mg/L for chloride-sulphate waters and 250–500 mg/L for bicarbonate waters, and that a bicarbonate content of 30 mg/L was the minimum “needed to achieve acceptable organoleptic characteristics” (Kozisek, in WHO, Nutrients in Drinking-water, 2005).

Read that as a distiller rather than a public health engineer and it says something useful: demineralised water sits deliberately outside the drinking-water palatability envelope. The mineral content that makes water agreeable to drink is the same content that would contribute its own signature to a reduced spirit — the mechanism set out in our piece on TDS, hardness and whisky’s nose.

Which is precisely why sensory testing matters more for purified water, not less. A mineral water carries a baseline character that masks small faults; strip it away and there is nothing to hide behind. Every trace organic, every migrated plasticiser, every note picked up from a tank or a hose stands alone against a blank background — the same asymmetry that runs through the microbiology of purified water and what storage and pipework do to a finished spec. Purity removes the buffer as well as the burden.

How should a distillery actually run this?

A defensible routine sensory check does not require a 22-person panel or seven months of training. It requires consistency, and about ten minutes per batch.

1. Sample at the point of use — where the water enters the reduction, not at the plant outlet. The distance between those two points is where most acquired faults live.
2. Assess cold and warmed. Follow IS 10500 and test both: warming to roughly 40–45 °C drives volatiles into the headspace and reveals notes that are silent at ambient.
3. Control the glassware. Odour-free, dedicated, rinsed with the water under test and never with detergent — a soap film will out-perform any contaminant you are looking for.
4. Control the assessors. No smoking, coffee, perfume or scented hand wash beforehand; a quiet, odour-neutral space. Two or three consistent people beat a rotating cast of ten.
5. Compare, don’t just sniff. A blind triangle test against the previous accepted batch is the highest-value check available, and it converts an impression into a pass/fail result.
6. Escalate with descriptors. Capture an FPA-style descriptor and intensity before re-testing. “Musty, weak” and “medicinal, moderate” point at completely different parts of the plant.
7. Test the water in the spirit. Reduce one control with your reference water and one with the candidate batch, then compare at nosing strength — conventionally the 20% ABV dilution used by tasting panels. This is the test that matches the use case.

None of this replaces the certificate. Conductivity, TDS, pH, TOC and microbial count remain the backbone of a blending water specification, and knowing how to read that spec sheet is still the first competence a buyer needs. Sensory testing covers the gap a certificate structurally cannot: the compounds nobody thought to assay, at concentrations nobody routinely measures.

Glossary

• Organoleptic — relating to properties perceived by the senses: odour, taste, colour, appearance. In IS 10500 these are the first parameters listed, ahead of any chemistry.
• TON (threshold odour number) — the dilution ratio at which a sample’s odour ceases to be perceptible against a reference water. Measured under EN 1622. A TON of 1 means the undiluted sample already has no detectable odour.
• TFN (threshold flavour number) — the same measurement made by mouth rather than nose.
• FPA (Flavor Profile Analysis) — Standard Method 2170. Four or five trained panellists describe and rate the intensity of a sample’s attributes without dilution, using reference materials to fix a shared vocabulary.
• Detection threshold — the concentration at which a stimulus is noticed as different from the blank. Lower than the recognition threshold, the concentration at which it can be named.
• Triangle test — three samples, two identical; the assessor identifies the odd one. A forced-choice discrimination test (ISO 4120) that yields an objective statistic from subjective perception.
• Reference water — the odour- and flavour-free water against which samples and dilutions are compared. Part of the method, not an incidental consumable.
• Geosmin — an earthy-smelling metabolite of cyanobacteria and actinomycetes, detectable at a few nanograms per litre and able to survive distillation and maturation.
• 2-MIB (2-methylisoborneol) — the musty-smelling companion compound to geosmin, with a comparable nanogram-per-litre threshold.
• Chlorophenols — reaction products of chlorine or chloramine with phenolic organics. Medicinal or antiseptic in character, with taste thresholds from 0.1 µg/L.

Frequently asked questions

Can a person really detect something at 4 nanograms per litre? A trained one, reliably. In the SUSPAW panel’s final sessions, 76.7% of assessors detected geosmin at 2.5 ng/L and 77.4% detected 2-MIB at 4 ng/L (SUSPAW). Training is what makes the difference — the same study screened 91 candidates and trained the survivors for seven months.

Should blending water be tasted, or is nosing enough? Both, where it is safe to do so. WHO’s thresholds show taste catching 2-chlorophenol at 0.1 µg/L against 10 µg/L by odour (WHO) — a fault that nosing alone will miss. IS 10500 attaches the obvious condition: taste testing is “to be conducted only after safety has been established.”

Does demineralised water have a taste of its own? It is engineered not to. WHO’s review notes water at 25–50 mg/L TDS was “described tasteless,” and recommends a 100 mg/L minimum for drinking water on palatability grounds (Kozisek, in WHO, Nutrients in Drinking-water). For spirit dilution that absence is the point — the water should contribute nothing but volume.

If my supplier sends a certificate of analysis, why test at all? Because a COA reports what was assayed. Conductivity, TDS, pH, TOC and plate counts say nothing about geosmin, 2-MIB or chlorophenols, and the concentrations at which those matter sit far below routine reporting limits. Sensory testing is the only screen that covers unspecified organics — and it is also the only check that catches a fault acquired after the certificate was issued, in a hose, a tank or a tanker.

How often should we run a check? Every batch or delivery, as a short-method screen against a retained reference — it takes minutes. Reserve full threshold determination or FPA for investigating a flagged sample, a new supplier, or a change of container or pipework.

Who should do it in a small distillery? Two or three consistent, non-smoking staff who are not otherwise involved in that day’s production, working blind against a reference. Consistency of people, glassware and reference water matters far more than the size of the group.

Blenders International · Water Science series